Laboratory Development anci Field Evaluation
of a Generic Method fol Santplirtg and Analy s i ~
of Zsocyanate~
Revised Final Reporl
Prepared for
Mr. Frank M'ilshirc National Exposure Research Laborator!
Air Measurements Research Divisiol hlethods Brand
US. Environmental Protect ion Agenc! Research Triangle Park, NC 2771:
Prepared by
J. F. McGaughe: S. C. Foste:
R. G. Merril
August 199:
CCADLAN C O R P O R A T I O N
EPA Report Number Month Pod Year of Publidion
Laboratory Development and Field Evaluation of a Generic
Method for sampling and Analysis of Isocyanates
Revised Final Report
Prepared by:
I. F . McGaughey S. C. Foster
R. G. Merrill Rac!im Corporation
P.O. Box 13000 Research Triangle Park, NC 27709
Project Officer
Mr. F r d Wilshire National E x p s w e Research L~horetory
Air Masurernents Researcb Division Methods Branch
US. Environmental Protection Agency Research Triangle P ~ r k , NC 2771 1
Disclaimer
?be information in this document has been funded wboUy by tbe United States Emironmedal Fjmtection Agency d e r EPA Contract Number 68-D1M)lO md 68-Dm22 lo Radian. It bPs been subjected to tbe Agemy's peer md dminirtrniive review, and it bas bsen approved for pblication rs an EPA document. Mention of trade names or commercial products does mt constitute dorsement or recommendation for use.
Acknowledgements
We wish to acknowledge the contributions of the following individruls to the rucccss of tbis program: Mark Owens, Danny Hurison, Durell Doerle, Jorn Runey, Pbyllk O'Hara, Jim Soutberlad, S a d Godfrcy, d Carol Hobson.
Abstract
Tbe U.S. Environmeatal Protection, d e r tbe .utbority of Title XU of tbe C h Air Act Amdmc& of 1990,
requires tbe identification d evaluation of r umpling d d y r ~ method for tbe following bocymates:
bexamethylene-1,6diisocymste; 2,4-tolueoe diirsocymste; meLhyleoe dipbeDyl diisocyraate; d matbyl isocyanate.
A laboratory rndy was performed to develop r sampling d d y r i s mabod. This effort wrs followed by three
fiekl validation tests using tbe approach specified in the EPA Metbod 301 protocol. Tbe sampling matbod employed
was r mcdification of the EPA Metbod 5, using r derivatizing agent to rtabilize tbe isocymstes for subsequent
d y s i s . Samples were d y z e d by high performance liquid chromatography d e r conditions sdquate to separate
PDd quautLfy all four co-. Tbe metbod bias d precision met the 301 criteria for an acceptable method si
tested at tbe specified source types.
Contents
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Section 1 Introduction 1-1
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Section 2 Conclusions IUKI Recomrnervlrrtions 2-1 2.1 Field Test #l (TDI) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-1 2.2 Field Test #2 (hlDI) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-1 2.3 Field Test #3 (HDI. hlDI. hll and TDI) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-2
Section 3 Experimenfal Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-1 3.1 Laboratory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-1 3.2 Field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-2
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Section 4 Results and Discussion 4-1 4.1 Laboratory Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-1
4.1.1 Literature Search and Background Information . . . . . . . . . . . . . . . . . . . . . . . . . 4-1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.2 Evaluation of Deriv~tizntion Techniques 4 4
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.3 Preparation of Derivatives 4-9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.4 Reaction Kinetics S t ~ ~ d y 4-15
4.1.5 Stability of Isocyanate Urea Cryshls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-16 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.6 Lnter-Lhoratory Round Robin 4-17
4.1.7 Lnstn~ment Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-18 4.1.8 Analysis of hll , HDI, TDI, MDI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-19 4.1.9 Instrument Daection Limit Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-23 4.1.10 Lnterfermts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-24 4.1.11 Toluene as an Lnlpurity . . . . . . . . . . . . . . : . . . . . . . . . . . . . . . . . . . . . . . . . 4-25
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.12 Train Spiking Experiments 4-28 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Field Evaluation 4-35
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.1 Quad-Probe 4-39 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.2 Quad-Train Assenlbly 4-39
4.2.3 Preparation for Smpling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 442 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.4 S m p l e Recovery 4-44
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.5 B 1 d s 445 4.2.6 Results of Field Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-46
. . . . . . . . . . . . . . . . . . . . . . . . . . . . Section 5 . Analytic~l hs tn~ment Performmce Rnd Quality Control 5-1 5.1 Overview of Dah Quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-2
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 Smlple Stornge d Holding Time 5 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3 Annlyticnl Qualily Control 5 4
Contents (continued)
Section 6 Metbod Quality Control d Metbod 301 Statistical Procsdurss ....................... 6-1 6.1 QualityCoatrol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-1 .
6.1.1 S.mplingQCROCddllrss ....................................... 6-1 ................................... 6.1.2 Iaboratory QAfQC Procedurss 6-3
6.2 Metbod 301 Statistical Procedures ...................................... 6-4 6.2.1 Cdculation of tbe Amount of Isocyanate Collected . . . . . . . . . . . . . . . . . . . . . . 6-4 6.2.2 NormaliLation of the Amount of Isocyanate CoUbcted . . . . . . . . . . . . . . . . . . . . 6-5 6.2.3 Precision of Spiked Samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.6 6.2.4 Precision of Unspiked Samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-7 6.2.5 Bias . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-7
Section 7 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-1
Figures
4-1 Stn~ctures of tbe icocy.nues. . . . . . . . . . . . . . . . .~...... . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-3
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-2 Reaction of szhPool witb 2.4.TDI 4-5
.......................................... 4-3 Reaction of 1. 2.pp with methyl isocyanate 4-5
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 4 Reaction of 2. 4.TDI with wrter 4 4
. . . . . . . . . . 4-5 HPLC chromatogram of underivatized 2. 4.TDI mobile phase 31:69 ACN:AAB (Isocratic) 4-10
4 4 HPLC chromatogram of 2. 4.TDI urea mobile phase . 31:69 ACN:AAB (isocratic) . . . . . . . . . . . . . . . 4-10
4-7 KPLC chromatopram of uoderivrtbed MDI Gradient Program . 31:69 to 90:lO (ACN:AAB) . . . . . . . . 4-11
4-8 HPLC chromatogram of MDI Urea Gredient Program . 31:69 to 90:lO (ACN:AAB) . . . . . . . . . . . . . . 4-11
4-9 HPLC chromatogram of underivntued HDI mobile phase . 31:69 ACN:AAB (Isocratic) . . . . . . . . . . . . 4-12
4-10 HPLC chromatogram of KDI urea Mobile Phase . 31:69 ACN:AAB (isocratic) . . . . . . . . . . . . . . . . . 4-12
. . . . . . . . . . . . . . . . . . . . 4-1 1 HPLC chromatogram of 1 . 2 . ~ ~ Mobile Phase 20:80 ACN:AAB (lsocratic) 4-13
4-12 HPLC chromatogram of MDI Uree Mobile Phase . 20:80 ACN:AAB (Isocratic) . . . . . . . . . . . . . . . . . 4-13
. . . . 4-13 HPLC chromatogram of 2. 4.TDI .Dd 2. 6.TDI (80:20) Mobile Phase . 31:69 ACN:AAB (ieocratic) 4-20
4-14 HPLC chromatogram of mixture of 2.6.TD1. 24.TDI. HDI. md MDI Grdient Program . 31:69 for 10 minutes to 5050 over 5 minutes . . . . . . . . . . . . . . . . . . . . . . 4-20
4-15 HPLC chromatogram of MI. 2.6.TDI. 2.4-TDI. HDI. d MDI gradient program . 20:80 for 4 minutes to 60:40 over 16 minutes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.22
4-16 Reaction of phosgene witb 1-(2-pyridyl) p i p e 6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-26
4-17 HPLC chromatopram of a high level soIution of phosgene derivrtized with 1. 2.pp . . . . . . . . . . . . . . . . 4-26
4-18 HPLC chromalogram of a low level sol ion of p b o s g ~ e derivrtiLad with 1 . 2 . ~ . . . . . . . . . . . . . . . . 4-27
4-19 HPLC chromatogram of a solution of tbe precipitate recovered from tbe derivrtization of phosgene with 1 . 2 9 ~ .................................................... 4.27
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-20 Spiking system configuration number one 4-29
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-21 Spiking system confipration number two 4-33
Figures (continued)
4-22 Spiking configuration number three .............................................. 4-34
4-23 HPLC chromatogram of contentb of first impinger . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.36
................................... 4-24 HPLC chromatogram of contents of second impinger 4.36
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-25 HPLC chrom~ogram of contents of third second impinger 4-37
4-26 Sampling train for isocyanates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-38
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-27 Qua~l-tmin probe and pitd arrangement 4 4 0
4-28 Upper and lower sampling probes (side view) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-41
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-29 Sample location for ? P I 4 4 8
4-30 sample location for MDI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-52
4-3 1 Schematic view of the sampling location for HDI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-57
Tables
1-1 tocygpplw Listed in the Clepn Air Act Amerdmeots of 1990 ............................. 1 . 1 . .
4-1 Physical Propertics of bocyrartw ............................................... 4-2
4-2 Reactivity Dais of 1socy.nate.s .................................................. 4-2
. . . . . . . . . 4-3 Recoveries for 2. 4.TDI Derivative Under V q i n g Conditions of Moisture. pH rod Solvent 4-16
4-5 Instrument Detection Limit d Limit of QlLantitation for 2.4.TDI. HDI d MDI Ureas . . . . . . . . . . . . 4-24
. . . . . . . . . . . . . . . . . . . . . . . . . . 4 4 Recoveries of 2. 4 TDI Using Spiking Configuration Number One 4-31
4-7 2. 4.TDI Breakthrough from First Impinger to Second Impinger Using Spiking Configur~tionNumberOoe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-31
4-8 Change in Impinger V o h e s from Start of Sampling to End of Sampling Using Spiking Configuration Number One . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-32
4-9 Recover). of 2. 4 J D I with High Moisture Using Spiking Configuralion Number Two . . . . . . . . . . . . . . 4-32
4-10 Recovery of 2. 4.TDI from Seven Replicate Spikes Using Spiking Configuration Number Three . . . . . . . 4-32
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-1 1 Summary of Method 301 Statistical Calcula~iots for TDI 4-47
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-12 2. 4.TDI Detected in Each Train. pg 4 4 9
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-13 Percent 2. 4.TDI Recovered from 2nd Impinger 4-49
4-14 Percent Recovery of tbe Spiked 2. 4.TDI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-50
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-15 Summary of Metbod 301 Slatistical Cdculations for MDI 4.53
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-16 Percent of MDI Recovered from 2" Impinger 4-54
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-17 Percent Recovery of tbe Spiked MDI 4-55
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-18 MDI Detected in Each Train. pg 4-55
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-19 Summary of 301 Statistical Cdculations for Twelve Rum 4-56
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-20 Percent MI Recovered From 2nd Impinger 4-58
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-21 Percent HDI Recovered From 2nd Impinger 4-59
Tables (continued)
.... .................................... Percent TDI Recovered From 2nd Impinger ; 4-59
........................................ Percent MDI Recovered From 2nd Impinger 440
Percent Recovery of tbe Spiked MI ............................................... 4-61
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Percent Recovery of tbe Spiked HDI 441
Percent Recovery of tbe Spiked TDI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 4 2
Percent Recovery of tbe Spiked MDI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 4 2
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . MIDetectedinEachTrain. pg 4 4 3
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . HDIDetectedinEachTrain. pg 443
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. 4.TDI Detected in Each Train. pg 444
MDIDetectedinEachTrain. pg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 464
Laboratory W t y Control Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-1
Data Qualjty Acceptance Criteria and Results Field Test #1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-2
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Data Quality Acceptance Criteria and Results Field Test #2 5-3
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . Data Qualjty Acceptance Criteria lad Results FieW Test #3 , 5-3
. . . . . . . . . . . . . . . . . . . . . . . . . . . Summary of Adyi ical N t y Control Results for Fiekl Test #1 5-5
. . . . . . . . . . . . . . . . . . . . . . . . . . . Summary of Analytical Q u d t y Control Results for Field Test #2 5-5
. . . . . . . . . . . . . . . . . . . . . . . . . . . Summary of Analytical W t y Control Results for Field Test #3 5-6
Section 1 Introduction
Sampling rod d y t i c a l metbods for r parti& d y t e or group of d y t e s cm be evaluated rod validated by
demonstrating their performance in fiekl tests, thereby ~lablishing the precision and bias of the methods
experimentally. Few mehcls have been fuUy evduted for sampling md d y z i n g the organic compounds listsd in
Title III of the Clem Air Act AmeDdments (CAAA) of 1990. For some analytes, methods have been developed UKI
evaluated for sample analysis, but not for sample collection. Field validation may be performed by side-by-side
comparison of r d i d a t e mdhod to r validated mctbod to eslablish comparable performance for the same d+ in
Lbe same matrix (same source category). Another procedure for validation of r mc&od is to spike known qunntities
of analytes into the collection appnratus in the field so that the precision and bias of the method can be demonhated
from sample collection through malysis.
The U.S. EnvironmenLal Protection Agency (EPA), under the authority of Title III of the CAAA of 1990,
requires the identification a d validation of sampling aDc1 analytical methods for the isocyanate compounds which are
listed among the 189 hazardous air pollutants identified in Title IIJ. These ibocyPnate compounds are listed in
Table 1-1. The laboratory developmental pbase of tbe method was accomplished under Work Assignments 11, 21,
d 40 on EPA Contract No. 68-D1-0010. The field evaluation of the method was accomplished under Work
Assijpnents 55 (TDI), 66 (hlDI), d 71 (HDI, MI, MDI, rod TDI).
In all, three field tests were performed to accomplish the isocyanate field evaluation.
Table 1-1. lsocyanates Listed in the Clean Air Act Amendments of 1990
Hexamethylene-1,6diisocyanate (HDII
2,4-Toluene diisocyanate (TDI)'
Methylene diphenyl diisocyanate (MDI)
Methyl isocyanate (MI1
The 2,6-TDI isomer may also be present but is not listed in the CAAA.
?be objective of this work was to develop d evaluate the isocyanate sampling sod analytical test method
through field testing at opernting skdiomq sources. The method was evaluated by collecting flue gas samples for the
analysis of the iodividual isocyanate d evaluating ibe data for bias rod precision. EPA Method 301, Field
VolidahbPt of Pollrcranr Measuremcnr Merhodrfrom V o h W p r ~ Media, was used rs tbe model for tbe experimotal
design of this mctbod evphration project. M y t e spiking was used with quadruplicate (QUAD) k p l i n g trains to
generate tbe required dab. Each fieM evaluation was performed .t m industrial facility which us& A rpscific
isocyanate in its manufacturing process. Only two of the quadruplicate tnins were spiked for each run. Tbe two
unspiked trains were used to sstablish tbe beck@ level of tbe target compound in tbe rt.ck grs.
?he umpling mstbod u t b a Mctbod 5- umpling train, which operates with 8 solution of I<-pyridyl)
piperazioe d toluene in tbe impingem. Stack gas is extracted from tbe Murce through 8 Wed, glass-lioed probe
and d r a m through tbe impingem. Any isocyrnste pmsd in tbe stack gas rssds rapidly with tbe pipe& to form
m ismymate derivative. ?he quantity of isocymrts M determined by first evaporating tbe impinger whrtion to
diybess d dissolving the derivatized isocyrnste in acdonitrile followed by high perfonaaoce liquid
chromatographic (HPLC) d y s i s .
This report discusses the d d s of the laboratory developmental work d tbe three fieM evphration studies.
Section 2.0, Conclusions d Recommendations, summarius the m l t s d provides recommePdations. Section 3.0
discusses tbe experimental approach for the laboratory d fieM rtudia d Section 4.0 provides a discussion of the
results. Tbe analytical instwent performance d rssocinied qual~ty wntrol data are provided in Section 5.0.
Section 6.0 presents the metbd qual~ty wntrol d EPA Method 301 statistical procedures.
Section 2.0 Conclusions and Recommendations
Based upon tbe nsults of tbe kbora!ory 8tudiw d three field tsrt evrhutions a! three different ~ w c e
cltegories, the following conclusions can be made oonoerning tbe validity of tbe method as tested under the conditions
described in this report:
2.1 Field Test #1 (TDI)
Tbe calculated values for precision (IRSD) for both the spiked Pod unspiked trPins, 3.55 Pod 4.72, respectively, are both well within the acceptance criteria of less than 5 0 1 RSD found in Section 6.3 of the EPA Metbod 301. Tberefore, the metbod as tested .t r flexible foam manufacturing facility meets tbe precision requiremsnts;
Tbe method bias, a! the concentrriion levels tested, was found to be significantly different from zero ri the 95 74 level of confidence when tbe dab from dl eigbt ~ K I S (minimum of six runs required) were used in tbe calculations. A correction factor of 1.053 would be required if d eigbt runs u e included in tbe bias calculation. Good technical reasons exist for excluding one of the eight QUAD runs. lf this run is statistically eliminated, w bins correction is required. In either case tbe correction factor is well within tbe acceptance criteria of 0.7 to 1.3 found in Section 1.2 of tbe EPA Method 301. Therefore, tbe method as tested af this source meets tbe bias and correction factor requirements; rod
Tbe metbod as tested is sufficiently robust to d o w testing sources similar to the source tested in this study wbere tbe stack gas moisture is less than 1 ?6 by volume, tbe stack temperahire is less than 30°C PDd tbe presence of otber c o r n p o d that may interfere with tbe analysis is m i d .
2.2 Field Test #2 (MDI)
The calculated values for prcxision (RRSD) for both tbe spiked and unspiked trains for eleven m, 4.25 d 28.2, respectively, u e both well within the rccepance criteria of less tban 50% RSD found in Section 6.3 of the EPA Method 301. Tberefore, tbe metbod u tested a! a pressed board manufacturing facility meets the precision requirements;
Tbe method bias, a! the concentration levels tested, was found to be rigdcantly different from zero .t the 95 % level of cod~dence wbert tbe data from dl eleven runs (minimum of six runs required) were used in tbe calculations. A correction factor of 1.084 would be required if .11 eleven ~ n s were ioctuded
- in tbe bias calculation. Tbe correction factor is well within tbe acceptance criteria of 0.7 to 1.3 f d in Section 1.2 of the EPA Method 301; Tberefore, the methul as tested af this source msets tbe bias and correction factor rquiresents; rod
?he mabod as Lestsd is ~fficiently robust to d o w tnting for MDI rr sarross simikr to tbe rource tested in this s a y w k r e the shck g.s moisbvs is k c than 5 96 by volume, the U~ck tempsmh~rs is k c than 45°C d tbe presence of ocber compoundr tba! may interfere with tbe d y c i r k minimpl.
2.3 Field Test #3 (HDI, MDI, MI and TDI)
Tbe dculnted values for precision (96RSD) for tbe rpiked trrrins for HDI, MDI, MI, uxl TDI, (3.3, 2.8,3.7, d 2.9, respectively) m dl well within the .coepCulce criteria of less than 50% RSD found in Section 6.3 of the EPA histbod 301. 'Ibedore, tbe mdbod as tested at r paint spray booth msstr tbe precision raquitemMts for r valid wlbod.
'Ibc mslbod bias for HDI uxl MI, at tbe coocentration kvek tasted, were found to be rignifimntly different from zero rt tbc 95 % confidence level when tbe d.tr from dl twelve runs (miPimum of rix required) were used in tbe dculations. A correction frdor of 0.9798 for HDI md 0.9701 for MI (rccepeble range 0.7 to 1.3) would be required if dl twelve runs m ioclrded in tbe bias crlculation. Tbe metbod bias for TDI and MDI were not ri@cmtly different from zero. Therefore, tbe mabod meets tbe bias criteria for a valid method for d four isocyanates as tested at r paint spray booth.
Tbe method rs tested is sufficiently robust to allow tc&ing at sources similar to the source tested in this sndy wbere tbe stack gas moisture is less than 1 A by volume, the stack temperature is less than 30°C d tbe presence of &r compouodb that may interfere with the d y s i s is minimnl.
Compile data and comments relative to testing at similar or different source categories in order to . further evaluatelrehe the method.
Document the sampling and analytical protocols following the Environmental Monitoring Management Council (EMMC) and CFR 40, Subpart 60 formats for similar sampling a d analytical methods.
Section 3 Experimental Approach
In order to ensure compliance with the C A M , it is oecess~ry to be able to accurately msrrmre tbe s m i r s h
of tbe listed isocyrnates from stPlionary sources. Sioce no vllijpted metbod(6) for measuring tbese compounds
existed at the time this work was begun, a laboratory study was initiated to develop a single sampling md analysis
procedure that would be amenable to all four isocyanales. Tbe source test method developed woukl be required to
satisfy the quality control criteria for method precision, bias, detection limit, md ~ggedness t y p i d y required for
source test methods. A method for measuring the emissions of isocyanates from stationary sources was developed in
a step wise approach.
3.1 Laboratory
The labor&ory development of the metbod followed the approach outlined below.
Perform a Literature search relative to the four isocyanates with emphasis on sampling a d analytical methds Pnd the chemistry of the compounds;
The literature review provided informdon on the isocyanates relative to their uses, physiological effects, reactivity and previous approaches to sampling and d y s i s reported by other investigators. Due to the reactive nature of the isocyaaates, two derivatization techniques were selected as possible approaches to developing a viable mechd.
Evaluate the most promising sampling and d y s i s appfo~cbes;
The fist approach tba! was evalunied was the reaction of isocylnates with a ~imple alcohol to form a stable urethane. No peak, were observed for the derivat id HDI rrd MI beuuse the compourwls have no chromophores detectable 254 nm using HPLC as the analytical technique. Since a single d y s i s technique for all four compunds was desirable, a second derivatiration technique was investigated. The use of I-(2-pyridyl) piperrazine (1.2-pp) p r o d u d derivatives of all four isocynnates with chromophores displaying suficient absorptivity to provide acceptable detection limits.
Perfom dynamic spiking of tbe cgmpouodb into a prototype urnpiing train;
- To simulate actual sampling conditions, a rpikiog apparatus was developed rrd used to dynamidy spike a proposed sampling train coaf;guration. Tbe results of this experiment were used to idedify and evaluate modlficatio~s to botb the sampling md d y s i s approaches.
3.2 Field Tbe mmlts of tbe kboratory experiwnts were used to formah the train configuration a d rrmple recovery
procedures for rctuol field evrhution of tbe metbod. Tbe procedures outlined in EPA Methods 1-5 wen iDcorporpted
dong with tbe d d y wnsideratioas necessary for using tohme .s rbe rolvmt in the impingers. It tben becune
obccssary to identify manufacturing facilities tht used one or mom of rbe four isocy.artes rod tbm obtain permission
to visit ench facility for tbe plrpose of wllecting umplss using cbs pro@ mdbod.
Section 4 Results and Discussion
4.1 Laboratory Development
The results obtained from tbe development of r cource campling method for tbe four isocyrnates listed in the
CAAA are discussed in tbe following sections.
4 . 7 . 7 L irerarure Search and Background lnformarion
Isocy~nates are used extensively in the production of polyurethnoe materials such as flexible foam, enamel
wire coatings, paint formulations, and in biders for tbe pressed boprd industry. Because of tbeir widespread use and
known adverse physiological effects such as bronchial asthma, dergic eczema, and skin irritation, several
isocyanates have been listed in Title Ill of tbe Clean Air Act Amendments. The isocyanates of interest include:
2,4-toluene diisocyanate P I ) ;
methylene diphenyl diisocymte (MDI);
bexamdhylene-l,6&socyanate (HDI); aod
methyl isocyanate (MI).
The aromatic isocyanares P I aod MDI) are used in the production of urethane foams utd pressed boards. The
aliphatic isocyanate, HDI, is used as a hardener in automobile spray paint .6d MI is used in organic synthesis and as
m additive to adhesives in polymer technology. Table 4-1 lists tbe pbys id properties, Table 4-2 lists the reactivity
&la, uxl Figure 4-1 sbows the structures of mcb of tbe isocyanates.' The isocyanate 2,6-TDI is no( included in Title
Ill of the Clean Air Act Amendments, but because it is present as r large percentage of the total isocyanates when ,
2.4-TDI is used it is included as part of the analysis, but excluded from the validation procedure.
Paramator HDI MDI M I
Aromatic lsocyanatr Aliphatic Isocyanate Chemical Family Chemical Name CAS Number Famula Ph y aical Form Color Odor MW B P C MPoC sol H,O
Arometic Isocyanate Aliphatic .Isocyanate
mrthyl k y u n l e 2.4 toluene diisocyanate hexamethylene-1.6- diisocyanate
12245-0
W a 4 Liquid
Wda4 Liquid
W , N O Liquid
Whict to Pale Yellow
h r p , Rr4pent 174.16
25 1 22 Not Soluble, R u c b Slowly virh Wakr
Cdorieu
h r p , Purrpent 168
213 -5 5 NM Soluble, R u c b with Water to Libcnc~ C 4
Whirc to Yellow No Odor 250.S 172 37 0.2g110Og Water at U)oC
Colorleu
h r p 57.1
39 -80 6.7111000 Water 2OoC, R u c b Slowly 0.96 at 25oC Specific
Gnvity Vapor
Table 4-2. Reactivity Data of lsocyanates
TDI HDI M D I M I Stability Stable Material at
Standard Temperature and Pressure
Stable at standard Temperature end Pressure
Stable at Stable i n solutions of acetone for Standerd 24 hours. Tempereture and Pressure No Informetion May occur at temperatures Available above 25.C
Hazardous Mey Occur; May occur Polymerization if in contact with
moisture or other materials which react with isocyanates. Self reaction mey occur at temperatures over 170°C or at lower temperatures i f sufficient time is involved
Incompatibilities Water, arnines, strong bases, alcohols, will cause some corrosion t o copper alloys end aluminum reacts with water to form heat, CO, end insoluble ureas.
May Occur; Contact with moisture or other materials which react with icocyenetes or temperatures over 204°C may cause polymqrization.
Strong dkelies, Water causes formation of CO, acids and alcohol. and methylamine gases. The
reaction with water is more rapid
Water, amines, strong bases. alcohols, metal compounds and surface active materids.
i n the presence of acids, alkalies and arnines. Contact with iron, tin, copper ( or salts of these elements). and with certain other catalysts ieuch as triphenylersenic oxide, triethyl phosphine and tributylin oxide may cause violent polymerization. Attacks some forms of plastics and ~ b b e r coatings.
Instability Temperetures Above High Heat and Tamperetures Moisture, mineral acids and Conditions 17O'C Moisture above 37.8.C bases Decomposition By high heat and fire: By high heat'and fire: By Fire: CO and CO, CO, and HCN Products CO. oxides of Nitrogen, CO,, CO, oxider of oxider of
traces of HCN, TDI Nitrogen. traces of nitropen. vapors end mist. ' HCN, HDI
LI
g::
NCO 2,4-Toluene diirocyanate
NCO
Methylene diyhenyl diisocyanate
Figure 4-1. Structures of the isocyanates and derivatizing agent.
Irocyrprtas undergo r e v e d different type6 of d o n s . The reaction of m isocyumte witb m dcobol
produces a ursthw. Tbe M i o n is a simple dcobol ddition mechnicm.2 The oxygen in tbe dcobol group
combioes with the &n on the c y m g m p of the isocyumte. Tbe displnced hydrogen from the dcobol combines
with the free electron on tbe nitrogen from the c y m group 14 form a stable urethoe. The uticle by E.H. Niemiw,
a d . , d d e d the potassium &oxide method for deriv.tizing uomrtic irocy.nrtcs witb dhawl @OH).' Figure 4-2
rbows tbe reaction of 2,CTDI with dbanol. 'Zhis reaction sbowed promise as r dcrivatiution d o n , W e the
reagents rre readily available aod the -ion rppsarsd 14 be immediate.
A recod reaction of isocymata is the reaction of m bocyumte with m amine which produces m uren. Tbe
reaction pmceds as a one-step ddition of the bocymate 14 the -NH, g m p of the d.' Tbe uticle by PA.
Gokiberg, a al., detailed a metbod for derivrtiring isocyanates with m aromatic h e 1-Q-pyridyl) piperazinc
(1 ,2-pp).4 Figure 4-3 sbows the ~ c t i o n of melhyl isocyanate with 1.2-pp. This methoJ was also promising rs a
derivstization reaction, because tbe reagMis are d d y available d aU isocyanates can be detected by ultraviolet
0 nbsorplion. A paper by V. Dharmarajan, d al., d d e d r method for generating isocyanate atmospheres,
sampling with mini impingers, and derivstizing tbe isocyanate with 1.2-pp.
Tbe third reaction bat an isocyanate can undergo is with water. Water combines with the isocyanate to form
an unstable ca&arnic acid (Figure 4-4). Tbe carbamic acid decomposes to form an amine and carbon dioxide, and in
the presence of an additional isocyanate, a stable polyurethane is formed. The reaction of diisocyanates with water is
tbe basis for the prductiou of polyuretbane foams. This reaction as an d y t i c a l tool is de s i r ab l e for d y s i s
because of the udmited number of polymers that could be formed in an uncontroUed environment (i.e., stack
emissions). The laboratory investigation of tbese reaction mechanism as possible approaches to tbe development of a
sampling and andysis method for stationary sources is discussed in tbe following sections.
4. 7.2 Evaluation of Derivatization Techniques
4.1.2.1 Potassium Ethoxide Method
The potassium etboxide derivatization method was investigated first because the available data showed that
there was an immediate reection with Ibe isocymales, the reagent was readily available, a d the derivatives were
stable for long periods of time. Tbe reaction of sthaool with an isocyanate is accomplished by a simple dcohol
addition mechanism (Figure 4-2). Tbe oxygen in tbe dcohol group from the slhanol combines with the &n on the
cyano grwp of the isocyanate. The displaced hydrogen from the alcohol combines with the free electron on the
nitrogen from the cyano group to form a stabk urethane.
0 I I
NCO E N - C O - C H
+ 2 040 - CH2 CH3) --> I Ethanol
NCO 6. EN- C O - C H ~ C H J
2,4-Toluene diisocyanate 2,4-Toluene diisocyrnate
Ethanol Urethane
Figure 4-2. Reaction of ethanol with 2,eTDI.
+ H3C-N=C=O :=>
Methylisocyanate 0 Q
I H
0 I C- N-CH3
1-(2-pyridyfjpiperazine / / I 0 H
Methylisocyanate Urea Derivative
Figure 4-3. Reaction of 1,2-PP with methyl isocyanate.
Tbe evlhution of the method procmdd by finl prspving the kocyaaab derivatives tbm analyzing by HPK.
Tbe deriv.titing mhrtion was prcprrsd by d i n g 1 p m of pDcurium hydroxide @OH) to 500 mL 99.9% dhuull.
Stock aohttions of each isocyrnrte wen prepued by dudving 10 mg of the iwcy.nrte in 50 mL m d h y h chloride.
A 100 pL .liquot of a h mlution w u dded to 20 mL of tbe hroiplion rolution d the pH djurtsd to 5 with
hydrochloric acid (37%). Tbe potassium chloride (KCI) precipitate was removed by filtration rPd washed with
ahmol, mviq the rinses. Tbe KC1 was discuded d tbe ~ p e - was evaporated to dryness .t 35'C uoder
v-. Tbe midue was dissolved in .I mL of (50:50) bthnno1:wnter d 20 to 50 pL was injected onto tbe column
of tbe HPLC.
Tbe derivativs wen d y d as fotlows:
Instnunent: Waters Model 5 10 HPLC;
Valve: Rbeodyne 6 port injector with 50 pL loop;
Detector: Vnrian W-50; 254 nm; 0.02 aufs;
Solvent: Acelonitrile (ACh3:Tetrahydrofuran flHF):KO (30:30:40): 1 mllmin; aod
Column: Zohax ODs C18 4.6 mm x 250 mm.
There were no peaks observed for HDI rrwl MI because they beve no strong chromapbores det4ctable at
254 nm. Several huge peaks were detected for 2,4-TDI d MDI. A positive identification of the derivatives, ideally
with the presence of single peak chromatograms, wns -say to evaluate tbe metbod.
The derivntization method wns, therefore, modified to d o w tbe otat isocyanate to be d e d directly to tbe
absorption solution. A 30 mg portion of MDI Md. 10 pL volumes of MI, HDI, Md. TDI were transferred to
iodividual vials containing 5 mL of tbe derivatiziog solution prepared as described above. To compare tbe derivatized
isocyanates with the derivatized isocyanates, a 30 mg portion of MDI d 10 PL volumes of MI, HDI, d TDI .
were transferred to dividual vials witbout tbe denvatiring reegenl d diluted witb h e m e . Encb of the solutions,
derivatized Md. underivatized, wns diluted 1:10 wilb tetrahydrofuran (THF) before d y s i s .
'Ibe following observations were made for tbe reaction of EtOH witb the isocy~nates:
TDI: Rapid evolution of gas and tbe formation of a white precipitpte.
MI: Rapid evolution of pas, w.prscipitpte formalion was observed.
- HDI: Evolution of gas wns wt observed. Two layers formed after the THF was dded d solidification occurred after 20 minutes. ?be rolid was dissolved in ACN for d y s i s .
MDI: Immediate reaction with slow evolution of gu, no precipilate observed.
h c b of tbe dutionr was d y r s d using tbe h v t HPLC oooditiolu. Tba chtomrtogrun of the derivatized
Underivntized TDI eluted a! 5.37 minutes rod derivatbd TDI autsd a! 4.71 mirum. Underivltitsd MDI eIutecl rt
6.86 minutes d derivatizsd MDI eluted at 6.36 mimh. No p d s were observed for MI or HDI, deriv.tited or
uoderivltized. Since MI d HDI coukl ~ o t be d y z e d by HPLC with W detection, gas chromatopphy (GC) with
A nitrogen specific detector (NPD) was employed rs m &mate mabod for d y s b of tbese deriv.lized icocy~o~tes.
Tbe deriv+iv,es were d y z a d on a GC under tbe following conditions:
Instrumeat: Vuian 6000 with ~ o u m p l e r ;
Detector: NPD, 320°C;
bjector: 320°C;
Column: RTX-5, 301x1 x 0.54 mm; aDd
Oven: 40"Cfor2min.,rampat10'C/min.to320"C,320"Cfor40min.
The chromatograms showed multiple peaks for each of the four derivative solutions rs well as the blank BOH.
Peaks occurring in the sample chromatogram bat were due to backg rod were eliminated by overhying the blank
chromatogram onto each sample chromatogram. Using Lhis approach, A major peak was identified for 2,4-TDI and
MI. No peaks specific for HDI aDd MDI could be identified.
Gas chrornatographiclmass spectral (GCIMS) d y s i s confirmed the presence of r TDI derivative d an MI
derivative. The other isocyanate derivatives could not be positively identified through GCfMS malysis. The
investigation of the potassium ethox.de reaction was suspencled because the HPLC analysis was selective only for the
aromatic isocyanates a d the GC metbod was inconclusive.
An altemtive approach was the reaction batween m isocyanate d the derivatizing compound 1-(2-pyridyl)
piperazine (1.2-pp).
4.1.2.2 1 -(2-Pyridyl) Piperazine Method
The derivatizing reagent 1,2-pp was cbosen because the aromatic ring can be detected by W spectroscopy.
With the addition of the iscxyanate, each compound could potentially be separated chromatographically by HPLC
with a W - detactor. Tbe d y t i d mctbod would Lben be responsive to each of the four isocyanates because the
derivatizing reagent would provide the chromopbore.
The derivahation reaction proceeds as a one-step addition of the isocyanate to the -NH group of 1,2-pp
(Figure 4-3). The analytical method d e s c n i in Reference 2 was evaluated as d e s c n i in h e following paragraphs.
4.1.2.3 Initial Preparation and Evaluation of 1.2-pp Isocyanate Derivatives Each u o c y W (0.8 mmol of xml nvted) was dhaolved ia r rsp.rrtc 10 mL v o h of mabylone chloride.
A 150 & v o h of Dut 1.2-pp wrr dded to -h icocyrnate dmdprd. Scpivltt solutions of mch of t k irocy.oltss
were prepated as above, ex+ Lhpt 1,2-pp wrr mt dded to the solutions. Each of the solutions wrr dilutul 1: 10
with metbyleae chloride. A 10 )rL aliquot wrr injected on the HPLC system dcrcnbdd below:
Instrument: Vuinn 5500 with urtoumpler;
Data System: Nelson 2600 (1 volt);
C o h : Hypenil BDS C18,S micron particle rim, 250 mm x 4.6 mm;
Mobile Pbe: Acebnitrile/O. lM Ammonium Acstrre Buffer (AAB);
For TDI and HDI: 31 :69 ACNI (0.1 M AAB, pH 6.2), irocra!ic;
For 1,2-pp md MI: 20:80 ACNI (0.1 M AAB, pH 6.2), icocratic;
For MDI: From 31:69 ACNI (0.1 M AAB, pH 6.2) to 90:10 ACN:AAB over 5 mins., gdient ;
Detector: Ultmviolet at 254 OLU (0.5 AUFS); lad
Flow M e : 2 mllmin.
Analysis showed that derivatives formed for dl four isocynmdes .ad Lhpt tbese components coukl be separated lad
d y z e d in the preseoce of reagents md unre~cted isocynmdes (Figures 4-5 thtougb 4-12). b e d on t h e Wings,
crystallioe derivatives were prepared, isolated, mJ purified for each of the isocynmtes.
4.1.3 Preparation of Derivatives
4.1.3.1 Dimethyl Sulfoxide as the Solvent (HDI, TDI, MDI)
The crystalline derivatives were prepared by first dissolving 0.8 mrnol of the isocyanate ia 3 mL of dimethyl
sulfonide (DMSO).' Then I ,2-pp (I .8 mmol) wns dissolvd in 3 d of DMSO. The two mixtures were combined,
beated to 60°C, lad stirred for 15 minutes. Tbe reaction was quenched by d i n g 200 d of deionized water, which
also uused the derivative to become insoluble md precipitate. Tbe precipitate was removed by filtration, washed
with water, .ad dried at 50°C. All of the isocynmdes fonned r precipitate exctpt MI. A separate approach w&
crtablished for MI lad is discussed in Section 4.1.3.4.
Ftgure 4-5. HPLC chrornatogram of underivatized 2,dTDI moble phase - 3159 ACN:AAB (Isocratic).
figure 4-6. HPLC chromatogram of 2,dTDI urea mobile phase - 31 :69 ACNAAB (iocratic).
4-10
MDI
Figure 4-7. HPLC chromatogram of underivatized MDI Gradient Program - 31:69 to 90:10
(ACN AAB)
hfDI Urea
Figure 4-8. HPLC chromatogram of MDI Urea Gradient Program - 31:69 to 90:10 (ACN:AAB).
Figure 4-9. HPLC chromatogram of underivatized HDI mobile phase - 31:69 ACN:AAB (Isocratic).
Figure 4-10. HPLC chromatogram of HDI urea Mobile Phase - 31:69 ACN:AAB (isocratic).
4-12
Figure 4-11. HPLC chromatogmm of 1,2-PP Mobile Phase - 20:80 ACN:AAB (Isocmtic).
Figure 4-12. HPLC chromatogram of MDI Urea Mobile Phase - 20:80 ACN:AAB (Isocratic).
4-13
Tbe HPLC d y s i s of tbe 2,4-TDI u r u sbowed several unidwtified peaks. Through a personal
communication with a mpmaeotative of a manufacturer of b o c y ~ , tbe extra peaks were ddennined to be tbe
result of a readion or aeries of reactions bd- tbe'solveot DMSO md tbe neat isocyanate. Tbe resulting solution
contPiaed isomers of tbe isocyade that contained oxygen substituents. Tbe use of DMSO was .baodoned in favor of
ACN, a less renctive solvent.
4.1.3.2 Acetonitrile as the Solvent (HDI, TDI, MDI)
Derivatives were prepared by dissolving 150 mg of each isocyanate in individual 5 mL v o h of ACN. Tbe
derivatizing solution was prepared by dissolving 300 IJ, of 1,2-pp in 5 mL ACN. Tbe two soiutiom w e n tben
mixed. A white p r e c i p i ~ e formed for each of tbe is0cyanat.a. The nrctions were qucncbd by adding 150 mL of
water. Each derivative was removed by filtration md air dried. Tbe prscipita&es w e n redissolved in a minimum
volume of ACN. A 150-mL volume of water was dded to precipitate tbe crystals, which were fdtered, air dried, md
transferred to 20 mL screw cap vials for subsequent analysis. Tbe HPLC analysis did not sbow my unidentified
pePks, rod tbe process for preparing the derivatives was considered a success. Tbere was a concern that unless the
ACN was dry, water may react with tbe isocyades fonning polymers. The use of ACN for tbe preparation of tbe
crystals was abandoned in favor of methylene chloride (M&g.
4.1.3.3, Methylene Chloride as the Solvent (HDI, TDI, MDI)
Derivatives were prepared by dissolving 500 mg of each isocyanate in individual 100 mL vohunes of M&J,
except for MDI which was dissolved in 250 mL of M C h . A 5-mL volume of 1,2-pp was dded to each of the
solutions, stirred rod allowed to reect overnight. A 150-mL vohune of hexane was added to each of the solutions to
precipitate tbe ureas. 'Tbe precipitate was then filtered md washed with 50 mL of hexane rod dissolved in a minimum
volume of methylene chloride. A 150-mL volume of hexane was aclded to precipitate the urea, followed by filtration
PDd a hexane wash. Tbe precipitate was again dissolved in MeCl,, precipihted rod washed with hexane, filtered, and
then dried at 50°C. This process produced white, powdered crystals which showed no impurities when analyzed by
HPLC .
4.1.3.4 Derivatization of Methyl Isocyanate
Special attention was paid to tbe derivatizalion of MI because it did nof form a precipitate wben added to a
solution containing 1,2-pp. The derivatiLation approach was modified to ve* tba MI could be derivatized and
analyzed along with the other isocyades.
Tbe MI derivative was prepared by transferring 100 pL of neat MI to 1 mL of ACN rod adding 300 pL of neat -
1,2-pp. Tbe solution was shaken for 5 minutes md then diluted 1: 1000 with ACN for analysis. 'This solution was
prepared so tbat the 1,2-pp would be in excess due to tbe toxicity of MI.
Tbe HPLC analytical conditions w e n chnged to: 20:80 ACN:Buffer for 4 min, r gradient from 4 m;nlrtrz to
20 min to 60:40 ACN:Mer . Two psllrs were observed in the chrommtognm, prsrumably tbe eicess 13% .rd the
&riv.tirsd MI. To dasrminc the idcDtity of t h e psllrs, a 20 fiL volume of MI WM then d d d to tb 1:lW
dihrtion rod tbe sohion d y z s d . Tbe 1,2-pp peak became smaller rod tbe MI p d became krger thus
wnfirming tbe f o ~ i o n of Ibe MI derivative .Id establishing the retention time of the MI derivative.
4.1.4 Reaction Kinetics Study
Tbe I ,2-pp was chosen for'tbe derivatiution of tbe irocyma!a b e w e 111 the isocymates could be
derivatired, chromatographically separated, d sbowed linear response with increasing concentration. A procedure
was tben developed to determine the completeness of the reaction between 1.2-pp and tbe ibocyrpptes.
Tbe complete~ess of Ibe reaction brd to be determined d e r varying conditions. The readion was studied
witb tbe &saxe of water, Ibe preseoce of W e r , tbe preseoce of acidic wrter, rod the presence of basic water. In
addition, toluene was selected as a candida!e for use in the sampling impingers because it is tbe solvent of choice for
many of the studies d e s c n i in the literature. Toluene has a lower vapor pressure than acetonitrile d wwld be
more Uely to be rerained in an impinger during sampling. Totuene is not miscible with water rod, therefore, the
possibihy of the reaction of TDI with water is minimized. Tohene is also less toxic than ACN.
A recovery study was performed in ACN md in toluene to determine the best solvent for the impinger
solution. Mixtures of known concentrations of TDI a d 1,2-pp were prepared in duplicate in ACN d in toluene.
The mixtures were prepared dry (no moisture added), witb 20% wrter, witb 202 water adjusted to pH 3 using
sulfuric acid, and with 20% water adjusted to pH 11.7 using sodium hydroxide. Standads prepared from the purified
crystals of the derivatued TDI were used to calibrate the HPLC. The solutions were analyzed d e r the following
HPLC conditions:
Instmment: Rainin HPXL Delivery system, Waters 710B WISP autosampler;
Data System: Nelson 2600 (1 volt);
Column: Zorbax ODs (4.6 mm ID x 25 cm);
Gradient: 25:75 ACNIO. 1 M Ammonium Acstate Buffer, pH 6.2, bold 2 min, then to 60:40 over 19.5 min, then resetting to initial conditions;
Ddector: Rainin Dynamax Dual-Wavelength, 254 nm (0.5 AUFS);
- Flow Rate: 2 mUmin; md
Injection Volume: 50 pL.
Table 4-3 NmmvwE rhc r ~ ~ ~ l t r of Ibc d y r i s . Tbc dtrivatimtion reaction bawsen 1.2-pp rod TDI indic~tes
completion for the ACN rolutions with md witbout water ddsd rr demonsMed by rscoveries of 90% or betier. The
game was tme for the toluene rolutions with md witbout water did. However, when acidic or h s i c water was
dded the per& recoveries from the ACN solutions drop to 0 percent. Tbe rscoveries from the tolueoe mixtures
with acidic or h i c water dded averaged 30% rod 8696, taspadively. Becurse then is a single phase rydem wben
ACN is the solvent, the aeot isocyanate M in conslant contact witb the acidic or basic water rod rhc d o n be(wee.n
1,2-pp md TDI is dversely affected. In contrast, wbea tohreoe in ucsd as the mlvsnt tbc isocyrrrPte is not in direct
cont~ct with the acidic or h i c water in a two-phase ryrtem md the derivatization M i o n is less affected.
Amount Target % Sample ID YQ YQ Target
ACN-TDI-(1-2 PP) 111 109 102
97 109 8 9
T OL-T Dl-(1-2 PP) 103 8 5 121
105 8 5 124
ACN-T Dl-(1-2 PPI-H,O 116 109 106
11 1 109 102
TOL-T Dl-(1 -2 PPI-H,O 96 68 141
95 68 140
ACN-T Dl-(I-2 PP)-pH3 0 109 0
0 109 0
T OL-T Dl-(1 -2 PP)-pH3 2 5 6 8 37
15 68 2 2
ACN-T Dl-(1 -2 PPkpH 11.7 0 230 0
0 230 0
TOL-T Dl-(1-2 PPI-pH 1 1.7 240 280 8 6
The studies described above iDdicate tbat the reaction between 1,2-pp and TDI does go to completion as
indicated by tbe high recoveries. However, the rate a! which this mction occurs was rtiu in question Pod, therefore,
a set of dynamic spiking experiments was designed to test the proposed sampling procedure d e r conditions ,$at
rimulakd actual source testing for isocyanates (Section 4.1.12).
4.7.5 Stability of Isocyanate Urea Crystals
'Ibe stability of tbe isocyanate urea crystals was evallrated or^ two separate occasions. Tbe stability of tbe
MDI-Urea crystals was evaluated in November 1993 during tbe secord field test, d the rtability of tbe HDI, TDI,
rod MDI-Urea crystals was evahrated in May 1994 during tbe third field test.
4.1.5.1 Stability of MDI-Urea Crystals (14 Months)
Ruifisd MDI-Urea cryst&, prspvsd Aupwt 22,1992, were wsd to prepuc tbe ctock colution used for tbe
second field tost ;piking d crLirrtioD solutions. Avifid MDI-Urea cryctds, prepared on November 11, 1993,
were used to evaluate tbe stability of tbe crystals prtparsd in August of 1992. Two cepuate working rt.ndard
soh&ns were prepared at .pp~oximatcly X ) p g / d d u) p g l d from each of tbe purified c r y d b U c k in ACN.
The solutions were d y z s d d tbe d t s r b o ~ e d LbPt ~h ~ h r t j o n frOm tbe 1992 cry& wu within 5 k of tbe
t u g a value. T k e drt. indicate tht tbe MDI-Urea cyrtrls ranah stable at k t 14 months wbm r t o d at room
temperature in r closed corJliner.
4.1.5.2 Stability of HDI, TDI and MDI-Urea Crystals (21 Months)
Purified HDI, TDI, MDI-Urea crystals (collectively as bourea crystals), prepared May 3, 1994, were used to
prepare tbe stock solution used for tbe third fiekl @st spiking d crlbration rohtions. The May 1994 batch of
bo-urea crystals was used to cbeck the stability of the August, 1992, batcb of bourse crystals. A five-point
ulibr~tion curve of tbe May, 1994, batch was prepared in tbe range of 10 pglmL to 50 pg/mL md d y M . The
correlation coefficient for each isocyanate was grater Lhn 0.99. Five s t u d a d s , 10 pg/mL to 50 pglmL were
prepared using the August. 1992, batcb of idourea crystals. The seodrvds were d y z s d md compared to the
calibrntion curve. Tbe results of the d y s i s sbowed that the HDI d MDIurea crystals were +5% of tbe target
value, while tbe TDI-Urea crystnls were -27% of tbe wet value. Tbe ruson for the low values obtained for the
TDI-Urea c&stals is that the d y s i s of the May, 1994, batcb did oot sbow a 2,6-TDI peak in tbe chromatogrun,
while the August, 1992, batch showed a 2,6-TDI peak that was 17% of the 2,4-TDI peak. The 2,4-TDI that was used
to prepare the August, 1992, batch consisted of 80% 2,4-TDI d 20% 2,6-TDI. The 2,4-TDI that was used to
prepare the May, 1993;'batcb was greater Lhn 98% pure. Using tbe 2,4-TDI calibration curve to crlculate the
amount of 2,6-TDI in urcb staodard and then correcting the mount of 2,4-TDI the average increases to -18% of the
get value. Since the actual purity of tbe 2,4-TDI crystals was oot determined, the actual stability of the crystals is
not known. However, there is no evidence to indicate that the 2,4-TDI crystals sbould not be as stable as the HDI
uxl MD1-urea crystals. These data indicate that tbe Iso-Urea cryrtals remain &le at least 2 1 months when stored at
room temperature in a closed container.
4.1.6 inter-L aboratory Round Robh
An inter-laboratory round robin d y s i s of lliquots of four of tbe ramples collected during the same field test
was o r g a d by Mr. Mark Baker of ICI Polyufetbanes Group. Seven Lboratories participated in the round robin
& M b g Radian Corporalion. Two umples inthe 1 pg/mL range, two ~l~ in the 10 pg/rnL m g e md r
1-2 PP/ACN blank were sent to eacb of tbe labor~tories for d y s i s . 'Ibe rample ideotificntions ue listed below.
The umples were shipped to tbe labrptories the week of November 14, 1993, rrwl the results were compiled by
Mark Barker a d made known to each labratory on December 20, 1993. Each labralory wns instructed to d y z e
(be samples using my metbud they selected. The results from urcb laboratory u e presented in Table 4 4 . The
results showed close agreement with !be average RSD (k) for d laboratories rod d analysis being 6.9 percent.
TsMe 44. Reaulta of Round Robin
Laboratory Biu ' Numbrt ualmL ( % I
1 1.46 12.4
Average 1.28 -0.8
Standard Dev. 0.10
Range 0.33
RSD (%) 8.1
Biu uglmL (%)
~dirn brmpk Numb
3
Uu uplmL (%I
8.84 16.6
4
Biu uglmL 1%)
8.03 12.6
7.91 0.8
'Uses Radian value as true value. % Bias = Laboratory Value - Radian ValuelAvg Value X 100.
4.1.7 Instrument Conditions
Derivatives were re-prepared by dissolving 150 mg of each ibocy.nate in individual 5 mL volumes of ACN.
The derivatizing solution was prepared by dissolving 300 pL of 1,2-pp in 5 mL ACN. Tbe two solutions were tben
mixed. A white precipitate formed for HDI, MDI, TDI. 'Ibe reactions were quenched by adding 150 mL of water.
Each derivative was removed by filtration, ah dried, and transferred to 20 mL screw cap vials for subsequent
analysis. Standards of each derivative were prepared at 80 pglmL in ACN. 'Ibe staadatds were analyzed using tbe
HPLC conditions below:
Instnunent: V k a n 5500 witb utosmqler;
Data System: Nelson 2600 (1 volt);
Column: Hypersil BDS C18 5p particle size, 250 mm x 4.6 tam;
Mobile Phase: Acetonit rile/O. 1 M AAB;
Detector: Ultraviolet af 254 nm (0.5 AUFS);
Flow Rate: 2 mYmin; UKI
Injection Volume: 10 pL.
' h e bocyWer 2,4-TDI md HDI w e n d y z s d i rocr r t idy al31:69 ACN:Mer. Figure 4-13 sbows a
npsreot.t ive chromatogrun for tbe d y s b of tbe two h r ~ of TDI. Tbt p d a u e g u r r s b in rbrpc with no
triling. MDI w u d y t e d wing a @ i d progrrm rtuting at 31:69 ACN:Ruffer for 10 minrltsr, t b ~ cbanging to
5050 ACN:Ruffer over the next five minrltar. A g r d i d was wed to keep the total run Lime u sboit u pwsible
(under 20 minuteb) .ad to rmidPin good peak sbrpc for MDI.
Next a single solution cont.ininp TDI, HDI, md MDI wrr prepand at 30 p g l d each in ACN. This rolutioa
was d y m d using the grdieat ptogrrm wed for MDI (Figure 4-14) to determine if all irocyanate derivatives
developed b this point cwld be d y d with one mdbod. The rdedion times listed below for tbe kocyrnrtes ue
specific to tbe H P K conditions givsn above:
2,6-TDI: 5.52 min;
HDI: 6.5 min;
2,4-TDI: 7.98 min; d
MDI: 17.45min.
Sufficient separation bdween the isocyrnates was observed to allow quantitative d y s i s . A three-point
dibration curve was prepared for 2,4-TDI, HDI, and MDI. A stock solution containing the three isocyrnates was
prepared by transferring 4.0 mg of each derivative to a single 25 mL vohunetric h k lrwl diluting to volume with
ACN. A three point curve was prepared by diluling aliquols of the stock to nominal concentrations of 8, 16, .ad 33
&mL of each isocyanate in ACN. The caLbration curve was analyzed using the above analytical conditions. The
slope, intercept, and correlation coefficient were calculated for each derivalive. The correlation coefficients were
0.999 or greater indicating that the analytical method is linear for three of the isocyaoates.
4.1.8 Analysis of Ml, HDl, TDl, MDl
A solution conlainiog 1.2-pp, MI, HDI, 2,4-TDI, lrwl MDI (each &a nomiaal concentration of 20 pglmL) wes
prepared Pod d y zed using tbe following instrument corditions:
Insttument: Varian 5500 with autosampler;
Figure 4-13. HPLC chromatogram of 2.4-TDI and 2.6-TDI (80:201 Mobile Phase - 31:69 ACN:AAB (isocratic)
MDI
Figure 4-14. HPLC chromatogram of mixture of 2,6-TDI, 2-4-TDI, HDI, and MDI Gradient Program - 31:69 for 10 minutes to 50:50 over 5 minutes.
Data System: Nelson 2600 (1 volt);
Cohunn: Hypcrsil BDS Cl8 5 micron p t i c b size, 250 rnm x 4.6 mm;
M i c a t : 20:80 ACN/(O. 1 M AAB, pH 6.2), bold 4 rnin tbso cbanging to @:40 for 16 min, tbso mdting to iaitipl coditions;
Dasctor: Ultraviolet at 254 nm (0.5 AUFS);
Flow Rate: 2 mUmine, rod
Injection Vohrme: 50 4.
Adequate sepadion between all components was observed for quantitrtion of each derivative (Figure 4-15).
Tbe first two of the three field tests evaluated tbe metbod using only om isocymate at a time (TDI d MDI,
~pcct ively) . Tbe third field test involved evaluating all four isocymateb of interest. The analytical mdbod used for
all four of the isocyanates as d e s c n i above is successful when the unount of excess 1,2-pp is kept at r minirmun or
MI is not r co rnpod of interest. Tbe amount of 1,2-pp in (be sampling train is at r concentration of 40 mg per
300 mL of impinger solution. Tbe results of (be first two field tests showed that r large 1,2-pp peak was present on
the chromatograms. There is baseline resolution between 1,2-pp md MI when (be concatration of 1,2-pp is less than
20 ppm. The coocentration of 1,2-pp in r sample with r 10 mL final volume is rpproximPtely 4 mg/mL which is
200 times higber tban (be minimum needed to obtain baseline resolution using the h o v e instrument conditions. A
column was purchased hat was more sLable with the buffer, md r guard column was added to extend the life of the
cohunn. The conditions that produced tbe best resolution Mween 1,2-pp md MI, while maintaining g o d reparotion
between the other isocy~nntes, were as follows:
Instrument: Varian 5500 with WISP Autosampler;
Data System: Turbochrom 4.0 (1 volt);
Column: Alltech Alltima C18 5 micron particle rue, 250 mm x 4.6 mm;
Grerlient: 1090 ACNl(O.1 M M, pH 6.2), changing to @:40 over 30 minutes, then resetting to init id conditions;
Detector: Ultraviolet a! 254 nm;
Flow Rate: 2 mUmin; md
Injection Volume: 50 pL.
Figure 4-15. HPLC chromatogram GI MI, 2,6TDI, 2,4-TDl, HDl, and MDI gradient progra- . x: ,~ , : for 4 minutes to 60:40 over 16 minutes.
1.2-pp: 4.3
MI: 10.2
HDI: 19.9
TDI: 21.1
MDI: 27.3
There was b a s e l k resolution between each of the irocyra~tes as well as tmtweeo 1,2-pp and MI, even with samples
similar to a 4 field wnples.
4.1.9 Instrument Detection Limit Study
The detection limit was ddermined for 2,4-TDI, I ,6-HDI, d MDI (MI was not under consideration ot this
time) using the Federal Register method for instrument ddsction limits.' A single staodard was prepared Chat
contained each of the isocyanates. The standard was subsequently d y z e d seven times on the following HPLC
system:
Instrument: Rainin HPXL Delivery system, Waters 710B WISP autosampler;
Data System: Nelson 2600 (1 volt);
Column: Zohax ODS (4.6 mm ID x 25 cm);
Gradient: 25:75 ACNIO. I M Ammonium Acetate Buffer, pH 6.2, hold 2 min, then to 60:40 over 19.5 min, tben resetting to initial conditions;
Detector: Rainin Dynamax Dual-Wavelength, 254 nm (0.5 AUFS);
Flow Rate: 2 mllmin; and
Injection Volume: 50 pL.
Tbe Instrument Ddection Limit (DL) was dculoted by determining the standard deviation of tbe uea wut
for seven replicate injections aIKI then multiplying tbe slandard deviation for seven replicate injections by the
Student's t-value ot the 991 confidence level (3.143). The Instrument Limit of Qunntitation eOQ) is estimated to be
3 times the D L (ng on columa).
The Estimated Method LOQ was calculated basal on a 10 mL final sample volume, a 50 p1 injection d a 1
cubic'mder sample coUected (Table 4-5). The IDL and the ILOQ are higher for HDI than for TDI or MDI because
HDI bas a lower extinction coefficient. This will be matrix .ad instrument dyredent .ad may vuy from laboratory
to Lbontory.
Table 4-5. Instrument Detection Limit a n d Limit of Ouantitation for 2,4-TDI, HDI a n d MDI Ureas s
1.6-HDI 2.4-TDI MDI
A u u s 7092 8204 20735
6930 8584 20504
6232 808 1 2022 1
6968 8474 20160
6787 8483 20132
6558 8477 2 1028
6098 7979 20600
Avcngc Area 6666 8326 20483
Auu Std. Dcv 382.8 234.9 334.3
Aru RSD 5.7% 2.8% 1.6%
Sundrrd UplrnL) 0.23 0.19- 0.22
ng on Column (50 pl injcc~ion) 11.5 6.0 1 1
IDL ng on column* 1.98 0.53 0.56
MDL nplrn' 396 106 112
ILOQ ng on column** 5.94 I .59 1.7
hf LOQ np lm'* 1 l R R 318 336
The Limit of Quantltation was calculated hased on a 10 n L final sample volume, and 35.314667 cubic feet ( 1 cubic meter) of air collected at sanq&ng tinie.
I ng isoqante injected .- IDLngonColunln =
A\.erage Area
* * I L O Q = I D L x 3 "* hlLOQ = MDL x 3
4. 7.70 interferants
- Area Std. De\,. Student's t value I
The identification of all ptential interferants is important but the process can he difficult and time consuming.
A source can have any nuinher of c o n p u n d s that could interfere with annlysis. Many interferants will not be knoun
before sampling and the type Rnd number of interferants will change from source to source. Two interferants,
aqueous solutions of strong acids and bases, have been discussed previously. Phosgene, which is a pas at room
temperature and is used in the production of isocyanates, has been identified as a third pss ih le interferant.
The popored reaction of p b o r g a ~ witb 1,2-pp ia r cubstitrrtion reaction wberc 1.2-pp r e p h tbe CI on
pborgene d 1.2-pp losu a H* (Figure 4-16). The rawtion occurs rt a 2:l motu d o consisting of 2 molu of
1 $-pp to 1 mole of pbosgcoe. Tbe formation of hydrochloric acid as put of tbe reaction mechaidm caJd rteo
prosent problems became the formation of tbe 1 . 2 ~ Locyraate urn is W e d by acidb.
Worgene was dsiivatizsd with 1.2-pp to determine if it wadd intsrfm chromrtogrrphidy with tbe d y r i s
of tbe Locyuutes. A 1 mL d q u d of r 20% mlution of pborgeae in t o h 11.93 M by volume] was transferred to r
100 mL h k co&g 500 & of 1,2-pp d 50 mL of ACN d brmgbt to vohune with tohreoc. A d
mhrtion wrr prepued by bansferring 500 & of a 20% dution of pborgene to r 100 mL flask wnr.ininp 500 pL of
1.2-pp d ACN d brougbt to volume with toh-. Tbe mlutions were filtered with a No. 42 Wbatmrn filter and
tbe precipitate iacovered .
Stock st.ndPrds of the prscipicpte were prepared at 50 pglmL in ACN d the stocks were dilutd 1 : 10 with
ACN UKI analyzed on the HPLC system d e s c n i previously. Figures 4-17 md 4-18 sbow chromatograms of the
hi& md low concentration staodards. The uea of a peak at 8 minuts decreases from the higb staorlard to the low
standard d was tentatively identified as the phosgene derivative. AU dher peaks were present in all of the
chromatogruns. The chromatogram of the crystals (Figure 4-19), does oot sbow my peaks that were od present in
Figures 4-17 md 4-18. The tentatively identified pbosgene derivative will interfere chromatographically witb the
d y s i s of 2.6-TD1 d 1.6-HDI under the current HPLC conditions. Since the presence of phosgene will probably
be known before sampling, the PaalyticPl corditions cm be cbanged to avoid tbis problem.
4.7. I I Toluene as an Impurity
During d y s i s of the samples an unidentified peak with a retention time of 19 minutes was observed d the
origin of the impurity was investigated. Three I5 mL volumes of HPLC g d e toluene were concentrated to dryness
with nitrogen at 6ScC aDd tben reconstituted witb 4 mL ACN for analysis. T h e 15 mL volumes of ACN were
prepared in tbe same manner as the toluene. Nea~ MDI was transferred to two separate I5 mL volumes of toluene
containiog I -0-pyridyl) piperarine aDd dowed to derivafize for 2 hours in'. sonicetor at 40°C. Each was
concentrated to dryness with nitrogen at 65°C md then reconstituted to 4 mL with ACN for d y s i s . Each r q l e
was d y z e d on the HPLC system previously d s s c n i . Tbese results sbowed oo peak at 19 minutes for my of the
samples that were prepared. A 20 uglmL solution of toluene in ACN was prepared aDd d y z e d . A peak at
19 minutes was observed. Tbe-se results sbow that tbe peak in question observed It 19 minutes was due to small
unounts of residual toluene remaining in tbe sample during preparation for d y s i s . This residual tolueae does DO^
interfere with tbe analysis, md, therefore, does not pose my problems in positively identifying the isocyanate
derivatives.
Phosgene
Phosgene Urea Derivative
Figure 4-16. Reaction of phosgene wtth 1-(2-pyridyl) piperazine.
Phosgene Urea
\
1 J PP
- r I I 1 I I .
1 I I I I
7 11 14 17 21 34 28 31 35
Figure 4-17. HPLC chromatogram of a high level solution of phosgene derivatized wtth 1,2-pp.
Figure 418. HPLC chromafognrn of a low level solution of phosgene derivatized wlth 1.2-pp.
Figure 419. HPLC chromatogram of a solution of the precipitate recovered horn the derivafization of phosgene with 1,2-pp.
4.7.72 Train Spiking Experiments
Trpin spiking experiments wen u n x b l a l 16 tsa the proposed mdbod in appropriate rpparrrtus under
Inborntory controlled d i t i o n s . Tbe spiking system used in this study was based on w o k plblisbed in r jamd
uricle entitled 'Recent Developments in Sampling rod Analysis of LocyrtrPtss in ~ i r . " ' Dynrmic spiking was done
to develop md test the possibility of using such r syrtem in tbe field rod 16 simulate suck ampling under corditions
~imilar 16 clean stack a~vironmeDtd.
4.1.12.1 Spiking Configuration Number One - Tbe spiking system, depicted in Figure 4-20, was designed to d o w dynamic #piking of the isocyrrrrtes, M
gases, into the rampling train. A motor driven syringe pump was used to deliver r solution of neat TDI in methylme
chloride nf r known flow rate into r beated (6OoC) nitrogen stream flowing It 3 literdmin. Tbe vaporized TDI was
h e n swept into tbe main heated nitrogen stream, flowing nf 15 Umin and into the rampling train. This system
dowed a constant rate of vaporized TDI to be carried into the rampling train. Tbe positive pressure of the spiking
apparatus transferred the entire amount of TDI in h e sy&ge into the train.
Tbe sampling train consisted of a glass transfer line (simulating tbe campling pro&) followed by r series of
five irnpinpers. Tbe first impinger (Greenburg -Smith) contained 300 mL of the absorbing solution, I ,2-pp, Tbe
remaining four impinpers were of the modified Greenburg-Smith design witb tbe seoord containing 200 mL of
absorbing solution, the third being empty, the fourth containing approximately 300 g of silica gel, sod the fifth
containing approximately 400 g of charcoal. A water jacketed cordenser was located between the outlet of the first
impinger and the inlet to the secod impinger to minimize the loss of toluene from tbe first impinger into the second.
No filter was used to avoid loss of isocyanate through retention on and reaction witb the glass filter matrix.
Using this system, five trains containing ACN and I ,2-pp and two trains using toluene and I ,2-pp as the
absorbing solution were spiked with known amounts of TDl. Eecb sampling ntn was nominally one b u r . Tbe fust
d second impinpers were analyzed separately to evaluate the ability ofithe solution to trap the isocyanate. Tbe
impinger solutions for each train were taken to dryness under vacuum nf 65°C and the residue then redissolved in 10
mL of ACN. The samples were analyzed under the following conditions:
Instrument: Rainin HPXL Delivery system, Waters 710B WISP autosampler;
Data System: Nelson 2600 (I volt);
- Column: Zorbax ODS (4.6 mm ID x 25 cm);
Figure 4-20. Spiking system configuration'number one.
Ordisot:25:75 ACNIO.1 M Ammonium Acetate Buffer, pH 6.2, bold 2 min, tben to 60:40 over 19.5 min, tbm msttiag to initid conditions;
Dscsctor: Roiain Dynrm~x Dual-Waveleqth, 254 nm (0.5 AUFS);
Injection Volume: 50 &.
Tbe TDI tecoveries averaged 74% (ranging from 59% to 80%) in ACN ud 88% (ranging from 86% to 91 %)
in toluene (7'able 46). Tbe average percMtage brocrkhrough from the first to tbe recod impinger WM 10% for the
ACN train ud 6% for tbe tobeae t& (Table 4-7). 'Ibe average percentage of rolvsd loss from the first ud m n d
impinges was 37% for the ACN train ud 23% for the toluene tnio (7'abIe 4-8). Taking into account the volatility of
ibe solvents, toluene was determined to be the k t cboice for the absorbing robtion ud all otber spikes were
performed using toluene.
4.1.12.2 Spiking Configuration Number Two
The system was next modified to more closely simulale actual campling conditions (Figure 4-21). A I-meter
glass probe beated to 120°C was installed bedweea the spiking system ud the impingem. In this configuration,
rample was pUed through the train using a staodard EPA Metbod 5 meterbox. Moisture was also introduced into the
rystem by applyiog beat to a glass reservou containing water connected in serias with tbe b a e d sampling probe.
Three 1-bow nuts were performed at a flow role of 18 Urnin through the impingers. An average of 100 mL of water
was collected in the impinpers representing 92.6 glm' of water (approximately 1456 by volume) within tbe simulnted
stack gas. The water was removed from the toluene absohing solution by transferring the sample to e 1000 mL
separatory funnel ilocl draining the water into a glass container, which was archived. The toluene absorbing solution
was transferred to a 500 mL r o d bottom flask, taken to dryness d e r vacuum ud then-reconstituted with 10 mL of
ACN in preparation for analysis. The samples were analyzed by HPLC to determine the average recovery. The
average recovery of the spiked TDI was found to be 642 (Table 4-9).
4.1.12.3 Spiking Configuration Number Three
The water bath was removed (Figure 4-22) to test the spiking system for precision Pod repeatability under the
lower humidity coditions (1-32 by volume) routinely found in the field. Seven trains were spiked with 2.5 pg of
2,4-TDI. Tbe contents of the fust Pnd secod impinpers were concentrated to dryness d e r a vacuum Pod
reconstituted to 10 mL with ACN. The peaks resulting from the aaalysis of the first d second impingem of each
train were combined for the evaluation of the recoveries. Tbe recoveries ranged from 24% to 122%. with an average
recoveryof 77% (Table 4-10). Tbe lmount collectd w e approrimately 23 times tbe instrument detection Limit Pod 7
times the limit of quantitation.
PQ Derivative
Train Total Target % ID 1st lrnpfnger 2nd lrnplnger Yg Pg Recovery
ACN
Tolurne
1
Average 3497 8 8
Table 4-7. 2.4-TDI Breakthrough from First lrnpinger to Second lrnpinger Using Spiking Configuration Number One
pg Derivative % of Total Collected
Train Total 1 st 2nd ID 1st lrnpinger 2nd lrnpinger Pg lrnpinger lrnpinger
ACN
1 1252 9 6 1348 93 7
Toluene
1
2
Table 4-8. Change i n lmplnger Volumes from Start of Sampling t o End of Sampling Using Splking Configuration Number One
Final Volume, mL
Train T o t d Starting % ID 1 st lmpinger 2nd Impinger mL Volume, mL Loss
ACN
Toluene
86 160
70 170
76 175
7 5 183
8 8 1 SO
Average
SD cv
Table 4-9. Recovery of 2,4-TDI wi th High Moisture Using Spiking Configuration Number Two
Train mL Air Vol Total Target % ID H,O L P9 Pg Recovery
Table 4-10. Recovery of 2,4-TDI from Seven Replicate Spikes Using Spiking Configuration Number Three
Total TDI Target Train ID bg)' b g ) % Recovery
1 3 -06 2.5 122
- Avenge 77 SD 38
. . cv 49 *
Absorbin Solution Empty Charcoal S~lica Gel Toluene i rperazrne
Figure 4-21. Spiking system configuration number two.
S ringe Jump
Nitrogen Gas @ 60' C
~ a s e r b i n ~ ~ o l u t i ~ n Empty Charcoal Silica Gel Toluene/ lperazlne
Temperature Sensors
Vacuum Gauge
Orifice I I
\i
Dry Gas Meter -
-
Figure 4-22. Spiking configuration number three.
Vacuum Lme
Figurn 4-23 through 4-25 sbow mpmcot.tive c h r o m r t o ~ of imp'iger conteatr from rpiked t d n s . 'Ibere
were r eved solveat implrities that did not interfere with tbc d y s i s of 2,4-lD1. ' h e impurities ue prsscat in
4.2 Field Evaluation
One of tbe objectives of this progrun WM to pcrfonn field tests to est&lish tbe bias rad precision of a
sampling IPd analysis mdbod for the isocyanate compounds listed in tbe C A M of 1990. Tbs bias a d precision of
tbe method were statistically determined following the protocol fivm io EPA Mabod 301. To achieve the test
objectives, mmufacauing facilities with b w n emissions of the specific isocyuurt6s were selected M field b s t sibs.
Flstors considered in the selection of the test sites were easy rccscs, ample space for tbe quadruple sampling ttrias
IPd proximity to Rdinn's ofice rad kborntory in Resauch Triangle Puk, Nortb Cuolinn.
Based on tbe results of samplcs collected during a preswey at erch boat facility, a sampling scheme wps
d c s i g d to sosure tbe collection of sufficient rrmplcs a d sample vohrw to yield statistidy valid data, IPd to
determine tbe level of the spike. Following the EPA Metbod XI1 protocol, quadruplicate (QUAD) trains were
opewed simultaneously with four co-located sampling probes. The first impinger of two of tbe trains were s t a t i dy
spiked with the derivotive(s) of the isocyanate(s) Pod two were unspikd. Somplas from tbe QUAD n m (minimum of
six valid ram required by Method 301) were returned to the laborafory and d y z d according to the analytical
procedures developed in the laboratory. These data were stalistidy evrhuted following Method 301 to determine
tbe performance of tbe metbd relative to bias and precision. The metbod is conriderad to be a valid metbod if tbe
precision criteria is met (% RSD <SO) a d tbe bias is not sipficantly different from zero or can be corrected with a
correction factor between 0.7 md 1.3.
Tbe sampling procedure used a modification of the EPA stationary source reference Metbod 5. Gas is
extracted from tbe source duct throuph a h d e d glass noulelprobe system as shown in Figure 4-26. The sample gas
stream is passed through a six impinger train. A water cooled condenser is placed between the outlet of the first
impinge1 a d the inlet of the second. Following tbe impingers, the gas is directed through a sample pump, a dry gas
meter and an orifice differential pressure meter. For tbe purpose of this work, tbe following modifidions were
made:
Four w-located sampling trains, referred to .s a 'quad-Lrain,' were used. Four independent samples were collected simultmeously uoder ideoticd stack coditions with the quad-train. This approach, dowed the collection of date to determine precision of the proposed sampling and analytical method.
1 J-PP 1
Solvent Impurities
Figure 4-23. HPLC chromatogram of contents of first impinger.
Solvent . Impurities
Figure 4-24. HPLC chromatogram of contents of second impinger, spiked train.
Figure 4-25. HPLC chromatogram of contents of third second impinger.
Temperature Sensor
Gooseneck : Nozzle
Heal Traced Quartz Probe
Temperature Indicator
Empty Charcoal s i l k
Temperature
Pump
Figure 4-26. Sampling train for isocyanates.
Tbe cootadr of the firct impiger in twm of each W of fau tmh wu spiked with a rtudud solution of one or more of the k y r s r t e c dcriv.ti;tsd with 1-Q-pyiidyI)pipm;tine. This rctivjty provided information on umpling d rPllytiCrl -v& d the bLr of the propored -ling d uulytid metbod.
4.2.1 Quad-Pmbe A specid probe assembly was required to d o w simultcroeous sampling from sssetltially the ume point with
four independent sampling trrins. V&ions in tbe velocity of the rtrrk gas within tbe uta ompied by the four
umpling mula must be minimized. 'Ibe quad-probe urrngemed dascribsd in EPA Mstbrd 301 mini- this
variation md has beeo p v m to be r good approximation of multiple-trrrio sampling from r ringle point.'
Figures 4-27 d 4-28 illustrate tbe configuration of the sampling probe which was used during the isocyanate
~ e s t program. The two primary considerations in the urrngement of tbe four sampling probes ue :
The probe inlets should be in the same plane perpendicular to the gas stream; and
Tbe probe tip openings sbouM be exposed to tbe same conditions.
Tbe flow at tbe probe tips cao be considered similar if tbe area eacompa~sed by the probe tip umnganent is
less than 5 56 of tbe stack cross-sectional uur.
4.2.2 Quad- Train Assembly
Four independent sampling trains make up the quad-train assembly. Although four meter boxes were required,
the velocity head (AP) was determined using only one of the four control boxes. The sampling trains were identified
as Train A, B, C, or D. Trains A d B were designated as 'spiked' trains for tbe fmt run. The spiking compounds
were statically spiked into h e first impinper of tbese'trains in tbe field for bias determination. The two spiked trains
for each run alternated between Trains A d B aDd Trains C rtwl D for d mbsaquent m.
Static spiking rather tban dy&c spiking of neat isocyanetes was performed in the field due to health md
d e t y cooceros.
upstream View (bottom)
k ~ r a i n D Train B ,
' Upstream Vlew (front) ~ r d ~ g d f o r d r u l l
Figure 4-27. Quad-train probe and pitot srrangment.
4-40
Figure 4-28. Upper and lower sampling probes (side view).
4-4 1
4.2.3 Preparation for Sampling
4.2.3.1 Glassware Preparation
All glassware used for rampling wps thoroughly c l d prior to use, iachxling the probe, impiigers, d
rample W l e s d d uteasils used during rample rscovmy. All glnsrware wns wnsbed with bot rorpy water, &ad
with bot tnp water, rinsed with distilled water d W e d in a o v a at 300°C for four bours. Tbe glrrsrware w u then
- triple rinsed with HPLC grade mxtoxitrile followed by triple riming with HPLC g d e tohrsoe.
4.2.3.2 Preparation of lmpinger Absorbing Solution
Data collected by Radian from p ' e h i n q ramples taken at tbe test rite facility was used b determine the
amount of tbe isocyanate Lhat wouM be collected in 30 cubic fed of shck gas. b e d on tbe stoichiometiic
relationship between tbe derivatizing &gent d tbe irocyarde oompound ddsctad (HDI), tbe minimum ~ m t r r r t i o n
of pipe& in toluene for use in the impingers was calculated. Tbe a c t d concentration of pipernine Dasds to be in
excess of this minimum amount to ensure Lhat tbe capacity of the absorbing solution is aot exceeded due to variations
in isocyanate concentration andfor sample volume collected. Therefore, tbe concentration of piperarine in toluene
was prepared at a level four times tbe calculated minimum d e d b e d upon tbe d y s i s of pre-survey ramples.
This solution was prepared in the laboratory just prior to going into the fieM and was used within 10 days of
preparption.
4.2 .3 .3 Preparation of Spiking Solutions
Sufficient spiking solution was dded to tbe first impinger so Lhat tbe amount present in the two spiked trains
was at least twice the amount expected in tbe two unspiked trains. The actual concentration of tbe spiking solution
was determined by tbe results of the prehnhvy sampling. Tbe spiking solution was spiked into tbe first impinger of
two of the trains prior to each quad run.
4.2.3.4 Sampling Equipment Preparation
The remaining pre-test preparation inclucled calibration and leak checking of all the train equipment, incMing
m d e h x e s , tbennocouples, nozzles, pita tubes, ud umbilicals. Refeience calbration procedures were followed
when available, and tbe results were properly documented ud retained. Ifa referenced catbration technique for a
particular piece of appanrtus was aot available, Cben a stale-of&-art technique was used. A discussion of the
techniques used to calibrate this equipment is presented below.
7. Tbe EPA A sspacied guidelines concerning the construction and porndry of
M acceptable S-Type pitot tube. Idonnation pertaining to tbe design ud construction of tbe Type4 pitd tube is
presented in detail in Section 3.1.1 of tbe EPA's Qualuy'Assurawz Handbodr for Air P o M m Meawrenvnr
System, Voktmt III, Sholvlry Source Specific M e t h d @PA 600f4-7742%). If tbe specified design and
construction guidelines are met, a pitd lube coefficient of 0.84 CM be wed. Only S-Type pitd tubes meeting the
-. GGkra mules were wed for sampling. AU m u l e s were tbomughly c l d ,
v W y inspected, .Id crkbrrtsd .ccording to tbe pmcahm cuUined in Sedion 3.4.2 of tbe EPA'r Dualih,
-.
Gas M&&U&m. Dry gas meters @GMs) were used in tbe umple tRins to msrnrrs tbe umple
vohune. Tbe sampling ntc w u also monitored; bowever, no exhaust gas vohrmetric flow rate was reported. All
DGMs were cdiirrtsd to doarment tbe vohuac correction factor prior to h. deputun of the equipment to tbe field.
Prior to d i r a f i o n , a positive predsure leak cbeck of tbe syrtem w . ~ performed using the pfocsdurs outlined
in Section 3.3.2 of the EPA's -. Tbe ryrtem was placed under approximately 10 inches
of water pressure rod r gauge oil mano-er was used to determine if r pressure decrease could be detecbd over r
one-minute period. If leaks wen delected, tbey wen eliminated before rcbal ulibntions wen performed.
AAer tbe sampling console was assembled aod leak checked, the pump was dowed to run for 15 minuted to
d o w Lbe pump rod DGM to warm up. The valve was Lben adjusted to obtain tbe desired flow m e . For the prc-tebt
crlirations, data were collected u the orifice manometer saings (AH) of 0.5, I .O, 1.5, 2.0, 3.0, rod 4.0 in 50. Gas vo- of 5 A' ut used for tbe two lower orifice oettiogs, rod vohrmes of 10 A' are used for tbe higher rsttings.
The iodividual gas meter correction fnctors (1,) are calculated for ~ c h orifice setting rod averaged. Tbe metbod
requires that each of the individual correction factors f d within f 2% of the average correction factor or the m-r
will be cleaned, adjusted, aod recalibrated . In addition, Radian requires thaf tbe average correction factor be 1.00
k l percent.
Dry gas meter calibrations were performed .t Radian's Perimeter Puk (PPK) lnborafory in Morrisville, N.C.
using an American wet test meter as M intermediate staodard. Tbe 'intermediale standard" is calibrated every six
months against Lbe EPA spirometer u EPA's Emission M w r e m e n f Lborrtory in Research Triangle P u k (RTP),
North Cuolina.
-of The four cample trains for each run were cbarged rod usembled in the remvery
M e r . The impinger buckets were clearly marked as Train A, B, C, or D. Tued impingers wen used.
Approximalely 300 mL of tbe absorbing reagent were t~nsferred to the first impbger d 200 mL to the record rod
third impingers. (Only two impingem with reagen! wen used in (be first field t a t for TDI.) Tbe first impinger was a
Greenburg-Smilb design rod d ltrmLiLLing impmgers were of tbe modified Greenburg-Smith design. ?he fourth
impinger tsMined empry, 200 to 300 g of silica gel was plncad in tbe fifib impioger, rod 400 g of chrcoal was
placed in tbe sixth impimger. A wuer jackded cordeaser wrs placed betw- the outld of the first impinger the
inlet to tbe r ~ n d impinger. Ten (10) mL of tbe spiking rohrtion was pipctted into tbe first impinger of Trains A .od
B. Openings were covered with Teflone film or d u ~ ~ foil after the u d l y of the t n h . '
Final assembly of tbe umple t& occurred at tbe umpling location. Thermocouples w m attubed to
-re the stack tsmperrrtrve d probe outlet ud impingar outk tamprrturss. Cnubed ice w u ddd b srcb
impinger buck&, rod tbe probe bsrters aYlDsd on ud dowed to strbiliza at 120" f 13 "C (248 " f 25 OF).
Tbe isocyanate tnins were I d c k k e d after umpling w u camplaad as required in EPA Mabod 5. LsPk
cbecks were also performed prior to umpling as a precaution~ry mawre. If a piem of g t s r w u e d e d b be
emptied or replaced, a final leak cbeck WM performed before tbe glassware piece was removed. An initial leak cbeck
was performed h e r the trPin was re-assembled.
To leak cbeck the assembled train, tbe m u l e end was capped off md a vacuum of 15 io. Hg was pllled in the
system. Once the vacuum bed been @led, the volume of gas flowing through the system was measured for
60 seconds. The leak rate is required to be less than 0.02 ocfm (ft'/min) or 4 % of tbe average r~mpling rate,
whichever is less. Afier the leak rate was determined, the cap was slowly removed from the noule e d mil the
vacuum dropped off, and then the pump was turned off. Iftbe leak rate requirement was not met, the train was
systematically cbecked until the leak was located and corrected.
Tbe leak rates and sampkng start and stop times were recorded on tbe sampling task log. Also, my dher
events that occurred during sampkng were recorded on the task log (such as pitd cleaning, thermocouple
malfuoctions, heater tualfunctions, and any other unusual occurre~ces).
Samphg train data were recorded every five minutes on standard data forms. With the single-pitot
arrangement used in the quad-test , the pitot tube was connected to only one of the four DGM boxes.
Recovery of tbe sample from tbe probes and mules was performed rt the sampling lockion. The sample
bottles containing tbe probe rod noule washings and sacb of the sampling trains were moved to the recovery trader.
Each impinger was carefully removed from tbe impinger bucket, tbe outside wiped dry, md tbe final impinger
weight determined and recorded. The isocyanate sample was then collected in tbe following fractions u reparate
samples:
First impinger contents, toluene rinses from the nodelprobe liner and toluenelacetonitrile rinses of the first impinger and connecting glassware; and
Conteats rrd tohrdrcs(onitri)e rinses from tbe #ecuod, third, rrd farrlb knpigers rrd tbe CODdeaser.
Rscovery procedures u e dotriled below. AU -very b o t h were wide mauth amber glass with Teflonm liosd lids.
4.2.4.1 Container 1 - Probe and First lmpinger Contents
?be cootsots of srch of tbe first impingem rrd first impinga connectors were iachded with tbe .ppropripta
p d d n o z z l e washing solution. 'Ibe -ire conteds of Lbc first knpiger wen recovered M r single w l e evm if
two p b c s were prasent. A small portion of tohreoe WM used to rinse tbe first impinger d connector h times.
A final rinse of tbe impinger with acedonitrile was rLo ~ e c s s u r y to rearove my water left on tbe k n p i e r w d d to
recover my remaining derivatizsd isocyraPtes.
4.2.4.2 Container 2 - Second, Third, and Fourth lmpinger ContentslCondenser Rinse
Tbe contents d tobenelrcetonitrile rinses of tbe second, third, md fourrb impingers rrd tbe condenser of
eacb trPin were collected in tbe same manner described h v e . Tbe contents of these impingers were d y z e d
separately from tbe contents collected in tbe first impinger to cbeck for breakthrough; therefore, care was taken to
avoid physical -over from tbe first impinger to tbe second. Any -over would invalidate tbe breakthrbugh
assessment and would be noted on the recovery documentation. Tbe contents of tbe fifth rrd sixth impingers were
weighed as previously described ad tben discanled.
4.2.4.3 Sample Storage and Shipping
Sample containers were checked to insure that complete labels had been affued. The labels identified
Trains A, B, C, or D as appropriale. Teflona lids were tightened rrd secured with Teflone tape. The sample bottles
were placed io a cooler, packed with ice, and returned to Radian's laboratory at the end of the field test.
4.2.5 Blanks
4.2.5.1 Field Blanks
Two sets of four complete trains were used for sampling during thiitest. Quad trains not in use for a
particular run were available for preparing field blanks. A total of four field blanks were prepared rod muwered.
The four field blPnks were prepared by syslematicnlly using T& A, B, C, rod D from each sst (A-1, B-2, C-1,
D-2). A field blank was prepared by assembling a sampling train in the staging uu, trking it to tbe umpling
location, rod performing a leakcback. Tbe probe of the blank train was heated during tbe sample test, but no
gaseous sample passed through the sampling train. Tbe umpling train was recovered in the same manner pnviously
described. -
4.2.5.2 Reagent Blanks
Aliquots of each lot of toluene, acetonitrile ad absorbing reagent were collected daily.
4.2.6 Results of Reld Tests
Three bost facilities were identified which agreed to d o w tbe use of their facility for the w.hution of the
method urxier rctupl test conditions. Each of b e facilities M discussed below in chronologicrl order, dong with &e
d t s of tbe field test.
4.2.6.1 Flexible Foam Manufacturing CTDI)
Vent gas samples conhining TDI were collected af a foam production pIpd 1 0 4 in Central NC during the
period February 22,1993 to February 26,1993. In tbe mLIlurnchrring procass stuting materirls WI, water, a
polyether resin, methylem chloride, an .mioe catalyst, and coloring additives) are blended in spscific proportions and
continuously fed onto a conveyer belt. 7be TDI reacts with H,O releasing C02 which curses the foaming action to
occur in the resin mer ia l . Mctbylene chloride (MeClJ can be added .r a supplemental 'blowing' or fcurming agent.
Tbe beat from the reaction of TDI and water muses the MeCL, to vaporize, resulting in increased foaming. 7be
degree of foaming (i.e., the density of the foam) is controlled by tbe mount of TDI and McCL, added. The foaming
action continues as tbe material procesds down tbe conveyer belt. Each continuous l a @ of finished product (or
'Bun'), is tben allowed to cure and degas for 24 to 48 bwrs.
Two production runs were performed each day Mocwlay through Friday, one in the morning and one in the
afternoon. Each continuous production run is one to two hours in length and may result in as many as ten different
types of foam based on foam density, hardness, color, and width. Tbe production of each foam type will consume
qproximately 15 minutes of the continuous production run. Tbe composition of the starling materials is changed to
produce the different types of foam. Tbe lower the density of a particular foam, the grurter tbe amount of TDI &or
MeCl, that is required in the starling formulation.
Three exhaust vents are used to remove the TDI and MeCL, vapors from over the production line.
Figure 4-29 presents a schematic view of the sampling location. Three induction (ID) fans u e used to exhaust
vapors from the production process through three separate uniasulated sbeet metal ducts that e x t d through the roof.
Two of tbe ducts u e comected by a 30' barizontal l a g & of duct, 34' in diameter, which then extends vertically to a
higbt of 25' above the roof top. A 6' diameter rampling port is located in tbe horizontal duct midway between tbe
vertical ducts RT lad #3, approltimately 5' feet above the roof level. Tbe roof level is approxirrmtely 15' above tbe
g d .
Eight QUAD runs were collected (two per day for four days). This number of ruas dowed for two extra rims
rbbve tbe required six in the event of sample loss during recovery, transport, or sample d y s i s .
Bias. Table 4-1 1 ia r summuy table which preieas tbe results of tbe rlltirtiul evaluation of tbe
test data following tbe EPA Metbod 301 criteria showing tbe metbod precision md bhs for 2-4 TDI. Metbod 301
q u i r e s vrlid drt. from a minirmvn of six QUAD runs. Table 4-1 1 peeeats drLa from rll eight runs. Prochion is
shown as the percent relative s t a d d deviation of tbe measured rmounts of TDI in the wnplcs. Rcsults for
precision of both tbe spiked d unspikd umples were kss tban 5 percent RSD, whicb ia well within tbe limits of
rccepcsble precision (upper limit of 50 k) givm in EPA Metbod 301.
Table 4-1 1. Summary of Method 301 Statistical Calculations for TDI
Parameter 7 Runs 8 Runs 7 Runs 8 Runs
Spiked Amount' 7828 7828 - - RSD, % 3.6 3.4 4.7 5.2
Average &as1 -295 -395 - -. Bias Significent? No Yes - - Correction Factor 1 .O 1.053 -- --
'Values are presented as p~ of underivatized TDI
Using the data from all eight QUAD runs, m&hd bias was measured at -395 micrograms. This value was
ddermioed to be statistical). significant at the 95 Z confidence level, using the t-statistic calculated for the d y t i c d
data. A correction factor of 1.053 was calculated for use with the melhod to compensate for the bias should the
method be used to measure TDI emissions from similar sources.
Using the data from only seven QUAD runs (elmhating run 8 because this run had the lowest average %
recovery a d the f d leak check for o w of the trains was questionable), the method bias was -295 micrograms. This
bias was not statistically sipdicant aod therefore no correction factor was calculated. In either case, the criteria for
.n acceptable methd were ma (i.e., a correction factor between 0.7 and 1.3).
6" Porl I\ 30' I
Figure 4-29. Sample location for TDI.
-. This section presents dscrik of tbe brsrbhroufi d m v e r y mu1tL for the
field crmplsr, u well as recovery infomution for tbe spiked coarpourd (2,CTDT).
m. Table 4-12 providbs r ~ m m a r y of the combined totd mrss of W T D I collected in the
probelfirrt impinger d recodthird impinger crmp1e.s for the uaspiked d rpikd trrinr, rsrpedively. Tbefe WAIL
.re ued as r h i s for cjculrting btsPkhtoufi of the TDI from the first impinger to the m o d impiier. Tbe mrsr
of oompouod found in tbe probdfirrt impinger f.iLction d in the mxmd i m p i e r f d o n was divided by the totd
mrss of 'I'D1 for chat trnin d then multiplied by 100 to yield the percat of the total TDI found in the repanate train
sections. Tbe d t s u e presented in Table 4-1 3.
Table 4-12. 2.4-TDI Detected in Each Train, yg
Quad Run Train 1 Train 2 Train 3 Train 4
1 12,263 1 1,520 4,735 4,811
2 6.287 6,270 13,258 13,733
3 11,625 12,078 3,466 3,166
4 2,132 2,139 9,694 9,805
5 12.233 12,717 4,947 5.522
6 8.042 7,608 15,741 15,010
7 10,450 9,290 2.558 2,616 8 4.086 4,579 11,197 10.770
%
Table 4-13. Percent 2.4-TDI Recovered from 2nd lmpinger
Train 1 Train 2 Train 3 Train 4 Qued Run % % % %
1 1.41 0.61 0.41 1.89
2 1.19 0.76 0.69 0.97
3 1.12 1.14 0.96 1.06
4 0.86 1.24 3.09 1.42
5 0.88 0.73 0.36 0.82
6 1.31 1.17 1.15 1 .29
7 1.99 1.68 1.07 1.62
Tbe average breakthrough for tbe spiked trains was 1.5 % 16 for tbe unspiked tnins 1.1 percent. Mom than
98 % of tbe TDI was collected in tbe first impinger under tbe campling conditions used in this d y .
Tbe rmount of tbe tolueae contained in tbe first impiigers of each of tbe trains wms reduced by .pproximrtely
25% (by weight) during tbe sampling run due to evaporation. ' h e second impingem rbowed, on tbe 8Vmge, r od
gain of rpproximrtely 5 percent. Tbe ranainder was d e c t a l in tbe #il ia gel mi tbe chtcorl , botb of which
rbowed I& weight g k . Tbe t d ~ I weight gained in tbe train components following tbe first impinger mom tban
compensates for lo- from tbe first impinger, probably due to the collection of a rmall uwunt of moisture. Tbe lors
of toheoe from tbe first impinger was minimid by keeping tbe impiierr in UI ice bath d pkcing a water cooled
condenser between tbe outla of tbe first impinger d tbe inla of tbe setand impinger.
-. Om of tbe objectives of this test program was to obtrin bias and precision data to validate tbe
proposed test metbod for isocyanates. Samples from two of tbe four trPins of cech quad assembly were spiked witb
TDI before cecb sampling run. The estimation of matbod bins is b e d on tbe percentage of tbe TDI spikes
recovered. Analytical resulb used for this calculation are tbe averages of the triplicate analysis results for each
spiked sample. A summary of tbe spiked TDI recovery percentages is presented in Table 4-14.
Table 4-14. Percent Recovery of the Spiked 2.4-TDI P
QUAD Run No. Train' 1 2 3 4 5 6 7 8
Average 9 5
% RSD 8 .2
'Spiked Trains alternated from run to run, A and 8, then C and D. 'Values are o percentdoe based upon a spiked amount of 7827.8 pg as 2 ,4TDI .
Tbe percent recovery was calculated for cech spiked train for each run foUowing the example calculation
procedures outlined in steps 1-3 of Section 6.0. The value obtained rd step 3, the amount of spike w v e r e d , is
divided by the actual amount of TDI spiked, 7827 pg, multiplied by 100. An average wovery was doterminal by
averaging tbe 16 individual run recoveries.
, Tbe recovery for TDI ranged between 83 md 112 percent and averaged 95 percent with a %RSD of 8.2,
4.2.6.2 Pressed Board Manufacturing (MDI)
v d g a -1- Cowg MDI w e n C O U ~ .t l prvrd M w f r c ~ u r i o g plrd w in rvarbm
Virginia duriog the period September 22, 1993 thra~gb September 26, 1993. kr tbe manufactwbg procuss, tbc boud
product is made from combining wood materials under pressure with an h i v e . Tbe dbesive formulation inchder
MDI url pbenyl fonnrldehyde.
Tbe process begins u whole logs ue dbbarked rod p s through a chipper, wbere tbe logs ue dud to
wood chip that ue approximately 4 square iaches. Tbe chip u e dried to a speciftc moisture conteat in a rotmy kiln
url tbea passed to a rotmy mixer where tbe birder -rial is dded. Tbe chip u e then fed onto a conveyor belt in
d o n s that u e approximately 4' x 8' x 3 '. Each section is then buted under hi& pressure for rpptoxknately
obe minute. Tbe resulting sbbets u e tben trimmed to siLe rod p h g e d for shipmeat.
Figure 4-30 presents a schematic view of tbe sampling locotion. Two 48-inch uial fans u e used to exhaust
vapors from tbe pressing process through a stack on tbe roof. Tbe inside diameter of tbe stack is 59.5 k b e s . There
is a stack divider in tbe section where tbe ducts from tbe two fans intersect the stack, which extends approximately 5
feet downstream past the junction. Samphg ports u e loafed approximately 20 feet downstream of tbe stack divider.
Tbe four 6-inch diameter sampling porn u e loafed equidistantly u d the stack, five feet above a circular
platform. Tbe platform is loaded approximately 38 fed above the roof rrd is accessible by ladder from the rooftop.
The roof level is approximntely 40 feet above tbe g r d .
A total of eleven quad runs were collected over the test period, allowing approximately 3 boun
turnaroud/recovery time between runs.
P r a . Table 4-15 is a summary table which presents the results of the statistical evaluation of the
test data foUowing t te EPA Method 301 criteria showing the methd precision aod bias for MDI. Metbod 301
requires valid data from a minimum of six QUAD runs. Table 4-15 presents data from all eleven mu. Precision is
s h o ~ n as the percent relative starwlard deviation of the mursurd amounts of MDI in the samples. The precision of
tbe spiked samples was less than 5 % RSD, d less than 30% RSD for tbe unspiked samples, which is within the
limits of acceplable precision (upper lrmit of 50%) given in EPA M M 301.
Using the daul from eleven QUAD runs, mdbod bias was measured r! -50.6 micrograms. This value was
determined to be slntisticdy sigrhcant at the 95 5( confdence level, using the 1-6tatistic cdculatd for tbe d y t i c a l
data. A correction factor of 1.084 was calculatkl for use with the method to compensate for the bias sbould the
mclbod be k e d to measure MD1 emissions from similar 6OUrCes.
Spiked Trains Unsplked Trains
Runs Runs Runs Runs Runs Runs Parameter 1 to4 ' 7t012 ' 1t012' 1 to 4' 7 to 12' 1 to 12'
Spiked Amount' 651 65 1 65 1 - - RSD. % 6.01 2.96 4.25 11.96 33.43 28.18
Avarrpe Bias' -30.7 -52.6 -50.6 - - - Bias Sipnificrnt? Yes Yes Yes - - - Correction Factor 1 .049 1.088 1.084 - *
'Samples were collected at a flow Rate of 0.75 cfm. 'Samples were collected at a Flow Rate of 0.60 cfm. 'Data ia a combination of samples collected at both flow rater. Run 5 was statistically excluded as an outlier.
'Values are presented as up of underivatized MDI.
Using tbe data from QUAD Runs I to 4 (Sampling flow rate of 0.75 cfm; icokinatic, d eliminating Run 5
because tbe data from Train A were statistically found to be outliers), metbod bias was measured d
-30.7 micrograms. This value was determined to be statistically signific~nt ol the 95% confidence level, using the
t-statistic calculated for tbe analytical data. A correction factor of 1.049 WAS calculated for use with tbe method to
cornpeasate for the bias sbould the metbad be used to measure MDI emissions from similar sources using a flow rate
of 0.75 cfm.
Using the data from QUAD Runs 7 to 12 (Sampling flow rate of 0.5 cfm; non-isokinetic, a d eliminating
Run 6 because the spike recoveries were sidcrotly lower ol 81 5% and 82 2 for the two spiked trains), method bias
was measured at -52.6 micrograms. This value was determined to be statistically s igdkant the 95 % confidence
level, using the t-statistic calculated for the analytical dala. A correction factor of 1.088 was calculated for use with
the method to compensate for the bias should the method be used to measure MDI emissions from similar sources
using a flow rate of 0.5 cfm. In m y use , the criteria for ro acceptable method were met (i.e., a correction factor
be.tween 0.7 suxl 1.3).
Recoverv.e extent of breakthrough of MDI from the first impinger to the secood during
sampling was determined by dividing the mount f o d in the secood impinger of each train by the total mou@ f o d
in the entire train less the mount due to the spike. Tbe results Ire presented in Table 4-16.
Train Quad
Averages
Run . A B C D Spiked1 Unspikedl All
1 4.2 2.7 8.2 2.6 a.6 6.4 4.4
2 2.9 10.4 12.0 28.7 20.4 6.7 13.6
3 7.5 7.1 0.3 10.4 7.3 8.4 7.8
4 6.8 18.2 10.4 8.6 8.5 12.0 10.8
6 5.9 6.1 11.6 7 .O 5.5 8.3 7.4
0 7 .O 8.1 12.2 8.3 10.3 ' 8 -0 8.1
7 4 .O 4.1 4.2 7.1 4 .O 6.6 4.9
8 3 .O 5.6 3.3 3.9 3.6 4.3 3.9
9 6.4 6.1 0.4 6.9 6.2 6.6 0.5
10 5.6 4.2 6.8 6.4 8.6 4.9 6.3
11 4.1 7.2 . 5.0 5 .O 5.6 5.0 6.3
'Spiked Trains alternate from run to run, A and 8, then C and D.
The average breakthrough for the spiked trains (Flow rate u 0.75 cfm Runs I through 5) was 9.2% with an
%RSD of 72, aDd 8.3% for the unspiked trains with a %RSD of 30. The average breakthrough for the spiked trains
(Flow rate at 0.5 cfm, Runs 6 througb 12) was 6 .01 with a %RSD of 37, aDd 5.9% for the unspiked trains with a
%RSD of 22. The average breakthrough for dl spiked trains was 7.3 76 with an XRSD of 63, md for all unspiked
trains 6.9% with m I R S D of 32.
O n e of the objectives of this test program was to obtain bias d precision data to validate tbe proposed test
mehod for isocyanates. Samples from two of the four trains of urcb quad assembly were spiked witb MDI before
urch sampling m. An estimation of method bias can be shown by percent recovery of the MDI spikes. Analytical
results used for this calculation are the averages of the triplide analysis nsults for each spiked sample. A cummary
of the spiked MDI recovery percentages is presented in Table 4-17 rod the total amount of MDI detscted is
surnmarized in Table 4-18.
Tbe recovery for MDI ranged b d w a n 55 and 105% for tbose trains collected u 0.75 cfm d averaged 91
perceot with an ZRSD of 15.3. The seven trains that bsd a collection flow m e of 0.5 cfm bed recoveries that ranged
between 81 and 101 A d averaged 91 1 witb m %RSD of 7.0. Tbe recovery for MDI for dl tmins collectrd
ranged from 55 to 105 '3 md averaged 91 % kith m %RSD of 11. -
T rain Quad Run A B C D
Table 4-18. MDI Detected in Each Train, pg - Quad Run Train 1 Train 2 Train 3 Train 4
1 775 744 113 119
2 63 7 7 751 664
3 627 698 8 1 87
4 9 9 127 717 754
5 48 3 735 146 111
6 117 122 650 656
7 807 768 223 152
8 254 122 739 728
9 669 622 64 7 2
10 145 145 ,775 805
11 746 722 122 133
12 64 67 689 674
4.2.6.3 Paint Spray Booth (HDI, MI, TDI, and MDI)
Veot gas samples containing HDI were coUected from the paint spraying operotion at m automobile
m a n u f a c t u ~ g facility located in tbe Eastern one-half of tbe United Stnles during tbe period May 24, 1994 through
May 27, 1994. In the manufacturing process automobile bumpers u e cprayed with m undercoating, a color coating
ud finally a clear top coating. Tbe p r m s is continuous from the time tbe untreated bumpers u e lorded onto
traveliug spray racks un~il they u e transported to tbe assembly u c a . Tbe first step is tbe application of tbe uodercoat - h e r which the bumpers pass through a drying oven before ibe remainiog two coats are applied. The bumpers then
From Spray Operation
- Damper
4" I.D. Port to be installed
/
4 Floor Level
-
) TO Incinerator
Figure 4-31. Schematic view of the sampling location for HDI.
4-56
Tbe plant operrtes three shifts per &y, Mordny througb F a y with scbeduld breaks during each rbift.
Sampling was comple4d prior to my b d .
EXhPllst vents from tbe spray b o b , ' f k b u)m' d drying ovens converge into a ringle duct tht lsds to
UI incioei.tor. Sampla were wllectsd from this dud. Figure 4-31 p s n t c a rcbcrmtic view oft& q k n g
location. Tbe uninsulated duct is 24 kba by 24 i n c h with r eved 2 inch porb located 107 inches above tbe floor
level. A new 4 inch pod was installed at the locltion indicated in Figure 4-31 by tbe facility prior to tbe rrmpling
date.
Biasand Table 4-19 is r summary table which praents the mulls of tbe statistical evaluation of the
t a t data following the EPA Method 301 criteria sbowing tbe method precision rod bias for HDI, MDI MI, rod TDI,
with HDI being the only isocyanate in the stack gas. hk4hod 301 requires valid drrla from a minimum of six QUAD
runs. Table 4-19 presents data from all twelve m. Precision is shown as the percent relative standard deviation of
the measured mounts of isocyanates in the samples. The precision of the spiked samples was less than 5 % RSD for
all four isocynnates, which is within the limits of acceptable precision (upper limit of 50%) given in EPA Method
301. The precision for the unspiked trains for HDI was less than 10% RSD, while the precision values for TDI rod
MI were less than 30% RSD. The precision for MDI in the unspiked trains was 56.1 % RSD. The poorer precision
values for the three compounds not present in the stack gas can be explained by the fact that the observed values are
most Uely due to trace conlaminalion that was detected near the detection limit where values are inherently less
precise.
Table 4-19. Summary of 301 Statistical Calculations for Twelve Runs
MI HDI TDI MDI
Parameter Spiked Unspikad Spiked Unsplked Spiked Unrpiked Spiked Unspiked
Spiked amount, pg 10.060 -- 161.4 .- 171.2 -- 164.9 -- RSD. % 3.7 -- 3.3 - 2 .O -- 2.8 -
- - - Average bias 304 3.325 1.788 -1.4 - Bas significant Yes - Yes - N o -- No - Correction factor 0 .9706 -- 0.9798 -. - -- - --
- Using (he data from twelve QUAD NIIS, &bod biss was rnessured at 3.33, -1.4, 304 .Dd 1.79 micrograms,
respectively, for HDI, MDI, MI UXJ TDI. The values for HDI rod MI were determined to be s tat is t idy rignir~cant
at the 95% confidence level, using the t-ctatis~ic calculated for (he analytical data. Therefore, a correction factor was
calculated for use with the method to compensate for the bias should the method be used to measure HDI or MI
emissions from rimilu w r c e s . Tbe crlcuhtd cmmction frcbr for HDI is 0.9798 and for MI is 0.9706. ' he biases
for MDI md TDI were mt significant md tberefon do mt require A correction frcbr. Tbe cribria for an .cospeble
mdbod were mst (i.e., A correction factor betweexi 0.7 d 1.3 and A I R S D <SO).
-. The exted of brsrldhrough of HDI from tbe first impiager lo tbe r d W
impier during sampling WM determined by dividing tbe total Mount f a d in tbe recondlthird tnpiogen of each
brin by tbe totd rmount found in tbe Wire train less tbe unounf due lo tbe spike. Any brrrhhtwgh for MDI, MI
and TDI wad13 oo( be due to sampling M tbese compauods were oo( in tbe stack gas but would be tbe d t of
impinger rolution carry over. Tbe results u e prsss~tcd in Table 4-20 through 4-23.
Table 4-20. Percent MI Recovered From 2nd lmpinger r
Methyl Isocyanate 2,606 p g Spiked
Train A Train B Train C Train D Quad Run % % % %
1 4 1 38 3 9 36
2 3 8 37 3 a 3 6
3 3 7 3 9 3 7 3 7
4 33 27 35 34
5 44 44 40 4 2
6 3 5 35 37 3 5
7 38 36 28 32
8 35 2 5 30 28
9 2 1 19 17 32
10 33 36 4 6 32
11 19 25 20 2 1
)(rxrmrthyl dikoeyrnrtr 161.4 w Spiked
. Treln A Train I Train C Treh D Qurd Run % 96 96 %
1 1.8 0 .O 0.4 0 .O
2 1.2 0 .O 0.3 1.1
3 0.3 0.8 0.7 1 .O
4 0 .o 0 .o 0 .o 0.B
5 0 .o 0.0 0 .o 0 .o 6 0.5 0 .O 0 .O 2.4
7 0.7 0.7 2.5 0.4
8 0.5 0 .O 0.3 0.0
B 0.0 0 .o 0.0 0.0
10 0.6 1.2 0.6 0.5
11 0.0 0 .o 0.0 0.0
Table 4-22. Percent TDI Recovered From 2nd lmpinger >
Hexmmethyl diimocymnmle 161.4 y g Spiked
Trrin A Train I Treln C Trrln D Quad Run % % % %
1 0.0 0 .o 0.0 0 .o 2 0.0 0 .o 0.0 0 .o 3 0.0 0.0 0.0 0.0
4 0.0 5.0 0 0 0.0
5 0.0 0.0 0 0 3.7
6 0.0 0.0 0 0 0.0
7 0.0 2.2 0.0 0.0
8 8 .O 0.0 0 .O 0.0 i
9 0.0 0.0 0.0 0.0
10 2.2 0.0 0.0 0 .o 11 4 .o 0 .o 0.0 0.0
1 2 17.6 15.6 0 0 0.0
Table 4-23. Percent MDl Recovered From 2nd lmpinger
M r t h y h o diphrnyl diiaooyanatr 161.4 pp Cplkrd
Troln A Train B Troln C Train D Qurd Run % % % %
1 ' 0 .O 0 .O 0.0 66.3
2 0 .o 0 .o 0.0 0.0
The average breakthrough for the spiked trains for HDI, MDI rod TDI was 1.3%. 0.38 5t and 0.262,
respectively d 3.3 '3, 2.3 'Z , rod 2.2 R for the unspiked trains. Tbe average 'apparent' breakthrough for MI was
34 percent. This 'apparent' breakthrough was due to an impurity present in the I ,2pp that presented a peak in the
chromatogram with a retention time similar to that of MI. Tbe uea of this peak coukl not be accurately removed as
backgrwod due to small variations in the volume of the I ,2pp solution d i e d to the sampling train impingers the
time of sample collection. Therefore, it can be assumed that if any breakthrough was present, the value would be
considerably less than the 30% reported.
One of the objectives of this test program was to obtain bias d precision data to validate the proposed test
mehod for isocyanates. Samples from two of the four trains of each quad assembly were spiked with HDI, MDI, MI
aod TDI before each sampling run. An estimation of methcrd bias can be shown by percent recovery of tbe t h e
spikes. Analytical results used for this dculation u e tbe averages of the triplicate Pnalysis results for each spiked
sample. A summary of the recovery percentages is presented in Tables 4-24 through 4-27. Tbe total pg i d in
each train is shown in Tables 4-28 through 4-31.
Tbe recoveries fcr all compounds ranged between 90 d 114 %. The average recovery for HDI, MDI, MI
aod TDI was 10074, W % , 103% and 101 %, respectively
Methy! booyonote 2606 yp lplked
Train A Train B Train C 7 rain D Quad Run % '16 96 %
1 103 104 0 0
2 0 0 104 103
Table 4-25. Percent Recovery of the Spiked HDI
Haxamrthyl Diirocyanala 161.4pg Splkad
Train A Train B Ttoin C Train D Quad Run % % % %
1 100 9 8 0 0
2 0 0 105 101
3 100 105 0 0
4 0 0 9 7 101
5 101 100 0 0
6 0 0 103 95
7 102 9 9 0 0
8 0 0 103 9 6
9 104 95 0 0
10 0 0 90 10 1
11 98 9 4 0 0
Table 4-26. Percent Recovery of the Sqiked TDI
Tdueru Diuocyrnato 171.2 UP Opiked
Train A Train B Train C Train D Quad Run % % % %
1 100 9 9 0 0
2 0 0 103 102
3 105 106 0 0
4 0 0 9 D 104
5 104 102 0 0
e 0 0 100 9 9
7 101 97 0 0
8 0 0 102 95
9 103 D6 0 0
10 0 0 9 6 103
11 100 97 0 0
Table 4-27. Percent Recovery of the Spiked MDI t
Mrttrylene diphenyl Diirocyrnrte 1 6 1 . 4 ~ ~ Spiked
Train A Train B Train C Train D
Quad Run Train A Train B Train C Train D
1 10,66.21 10,721.36 296.98 269.33
2 303.68 311.76 10,783.20 10,818.89
3 1 0,89 1.70 10,962.76 a16.13 288.66
4 a41.25 295.1 1 11.093.66 B359.34
S 10,524.07 10,936.88 286.83 217.34
6 310.75 322.03 10.899.79 9,920.60
7 10,752.78 10,754.89 275.15 343.84
8 291.83 428.84 10,716.60 10,642.57
9 10,672.83 1 1,057.59 673.41 373 -07
10 390.77 327.61 9,908.45 10,699.38
1 1 10,730.82 10,562.45 474.91 410.60
Table 4-29. HDI Detected in Each Train, pg
Ouad Run Train A Trrln 6 Train C Train D
1 175.00 170.94 13.53 12.64
2 66.71 66.68 236.52 230.3 1
3 248.75 257.56 88.66 86.53
4 10.22 11.52 166.99 173.57
5 186.60 183.77 23.72 21.94
6 48.19 49.54 214.69 203.17
7 205.07 199.93 42.29 38.73
8 15.22 15.83 182.29 169.65
9 178.03 164.32 12.43 8.95
10 121.23 118.13 264.69 282.41
1 1 184.28 178.05 26.74 2 6 .OO
12 8.50 20.73 190f74 195.20 s
Table 4 - 3 a g g
Ouad Run Train A Train B Train C Train D
1 172.36 171.24 1.70 2.25
2 2.01 2.17 178.81 176.33
3 181.01 182.69 2.11 1.94
4 0.64 0.97 169.61 177.98
6 180.38 177.61 2.49 2.1 6
6 1.32 1.76 172.67 171.04
7 174.20 168.1 1 2.64 1.63
8 2.09 1 SO 176.04 165.39
9 178.36 166.53 2.73 0.65
10 2.15 2.98 1 67.64 178.66
1 1 173.98 169.15 2.85 2.55
12 1.46 2.01 180.37 187.51 b
Table 4-31. M D I Detected in Each Train, yg
Quad Run Ttaln A Traln B Traln C Traln D
1 161.81 161.69 2.72 4.1 5
2 0.00 0.00 169.21 166.94
3 169.09 170.48 0.00 0.69
4 0.00 0.00 158.84 166.60
5 166.24 164.28 1.22 0.00
6 0.26 0.00 169;25 160.38
7 165.98 160.55 0.08 1.87
8 0.04 0.00 167.46 157.68
9 168.34 156.70 1.75 0.00
10 1.93 1.32 161.21 168.25
1 1 162.60 160.75 3 -68 5.05
12 5.07 5.18 169.67 176.1 4 b
Section 5 Analytical Instrument Performance and Quality Control
Tbe laboratory QA progrun for this project inchdcs proper handling, logging d tracking of incoming
samples, procedure vPlidations inchding HPLC column efficiency, calibration curvcs, daily QC cbecks d replicate
d y s e s , rod collection d o r d y s i s of sample, field rod reagent blanks, rod metbod spikes as well as field rod
laboratory spikes. A summary of the laboratory QC procedura detailed in this section is provided in Table 5-1.
Table 5-1. Laboratory Quality Control Procedures
Quality Analytical Control Acceptance Corrective
Parameter Method Check Frequency Criteria Action
Linearity Check
HPLC Run 6-point At retup or Corral, coeff. Check integ., reinteg. If curve when check std. >0.995 nacessary, racalibrate
is out-of-range
Retention HPLC Anal-yre Check 116-8 injections i 15% day-to- Check inotr. funct. for Time Standard day: i 5 % plug, ate. Heat column;
within one day Adjust gradient
Calibration HPLC Analyze Check 116-8 injections i 10% of Check intag., remake Check Standard rnin. 21cet calibration curve ctd. or racalib.
System Blank HPLC Analyze 1 /day Acrtonitrile
sO.1 leval of Locate courca of expected contam.: raanalyre analyte
Replicate HPLC Re-inject 1/10 samples or i 10% of first Check intsg., check Analyses Sample l l s e t inj. ' instr. function,
reanalyze
Method HPLC Analyze Spikes 1110 samples i 20% Spikes
Check intag. check inotr. function, reanalyze, re-prepare if
AS a parc of the testing for Work Assignments No. 55, 66, rod 71, Radian designed rod implemented a qual~ty
rssurance/&ty control (QNQC) effort tailored to med the specific needs of this project. ?be testing was
coducted in accordance with QNQC procedures descnhd in the Qualrty Assurance Project Plan (QAPjP) prepared
for each fieW test. Tbe results of the QNQC efTolr demonstrate that the dah u e reliable rod meet project objectives
for complaeness rod representativeness. Tbe datP m a the QA objectives for precision rod accuracy rod there are no
data quality issues that effect conclusions regarding tbe objectives of this project.
A r u ~ ~ ~ a r y of d y s i s results for QNQC samples, which includes measures of precision md accuracy .od
limitations in tbe use of Lbtse dah is preseded in this section.
5.1 Overview of Data Quality
7bc QAPjP estabhbed specific QA objectives for precision (15 96 .RSD), .ecurocy (f 30%). md comp1ete~e.s~
( 1 0 % ) for tbe detemb!ion of isocyrnate emissions. Tbe dptkticd results pmsntsd in Tables 4-1 1,415 md 4-19
md the % recovery v&es givm in T&les 4-14.4-17 md 4-24 thmugh 4-27 fibow tbat tbe objectives were met. 'Ibe
data @ty mxptmw criteria md tbe experimental d t s for each field test m summarized in Tables 5-2 through
54. Results for spike/spike duplicptes md triplicate d y s e s Gre compared with tbe criteria. In rll crisss tbe
criteria were met. Other dah quality indicators for each type of d y s i s m rlso prescded throughout the' rsmainder of this section.
Tbere ue w c a m wbere data @ty issues impair tbe study's conclusions with respect to tbe validity of the
sampling md d y t i c a l test method procedures. With exception of r limited number of samples, the @ty of
measurement dab generated for the test parameters fully meets tbe data @ty objectives outlined in the QAPjP.
Table 5-2. Data Quality Acceptance Criteria end Results Field Test X 1
Parameter Criteria Results
TDI Spike Recovery 70 - 130% 83 - 112%
TDI Analysis Results QUAD Train, % RSD 15% 5%
Individual DGM Correction Factor Agreement i 2% of Avg <2%
Analytical Balance r0.1 g of Class S Weights <O.l g
HPLC Linearity Correlation Coeff. >0.995 0.8995
HPLC Retention Time Variation f 15% f 10%
HPLC Calibrstion Check f 10% of Curve f9%
HPLC System Blank <0.1% Analyte level <0.1%
HPLC Replicate Analyses f 10% of 1 st injection i 2%
HPLC Method Spikes f 20% of theoretical f5% . %
Paremeter Criteria Results
MDI Spike Recovery 70 - 130% 82 105%
MDI Arulysis Results QUAD Train. % RSD 15% 7.9%
Analytic d b l a n o e r0.1 g of Clara S Weights <0.1 Q
HPLC Unearity Correlation Coeff. >0.995 0.9995
HPLC Retention l i m e Variation f 16% f 1.2%
HPLC Cdibration Check f 10% of Cuwe f 8%
HPLC System Blank <0.1% Andyte level <0.1%
HPLC Replicate Analyses f 10% of let injection 2%
HPLC Method Spikes i 20% of theoreticel f 5%
Parameter Criteria Results
HDI Spike Recovery 70 - 130% SO- 114%
MDI Spike Recovary
M I Spike Recovery
TDl Spike Recovery
HDI Analysis Results. %RSD
MDI Analysis Results, -%US0
M I Analysis Results, %RSD
TDI Analysis Results, %RSD
Anelyricel Bdence
HPLC Linearity Correlation Coeff.
HPLC Retention Time Variation
HPLC Calibration Check
HPLC System Blank
HPLC Replicate Anelyses
HPLC Method Spikes
15%
15%
15%
$0.1 g of Class S Weights
>0.995
f 15%
* 10% of Curve
<0.1% Analyre level
i 10% of 1st injaction
i 20% of theoretical
5.2 Sample Storage and Holding Time
Sample bold times apscified in the QAPjP were md for d umples. AU samples were prcparsd within
30 days of collection and analyzed within 30 days of prepration.
5.3 Analytical Quality Control
R d t s for metbod rpikes, field rpikes, field blanks, rsrgenl blanks and metbod blanks rre ad in
Table 5-5 through 5-7. Tbese samples sewed tbe dual pvpose of controlling and assessing meesunment drl. quality,
d providing tbe basis for precision md accurncy estimates. The QC acceptance criteria for each of h e types of
umples were met as sbown in Table 5-2 thtougb 5-4. Fiekl blanks were callected by assembling a umpling tnin as
if to collect a umple, transporting to the umpling l ~ i o n , leak cbeckhg md retuning the train to the onsite
laboratory for recovery. Field spikes and mstbod spikes were prepared by spikhg approximately 300 mL of the
toluene 1,2-pp solution witb 15 mL of the field spiking solution. Metbod and reegent blanks were prepared by
evaporating approximately 300 mL of solvent to dryness d dissolving any residue with 10 mL of ACN. AU spikes
and blanks mat tbe data acceptance criteria listed in Table 5-5.
No blank contaminarion problems were identified during tbe analysis of field atd laboratory blanks atd w
blank corrections were performed for tbe reported data. AU blank d y s i s data are presented dong witb other QC
and field sample results in Appendices A, B, and C.
Total Detoctod Theoretical Porcont Recovery Sample ID YP UP %
F i M Blank A 2.6 NA' N A
field Blank 8 8 .3 N A N A
Rald Blank C 6.4 N A N A
Rald Spike 1 7670 7828 06.7
f ia ld Spike 2 7686 7828 08.2
Mothod Spike 1 8120 7828 104
Method Spike 2 7838 7828 100
Method Spike 3 7890 7828 101
Method Spike 4 7945 7828 101
Toluene Reagent Blank' 0.5 N A N A
ACN Reagent Blank' 0.4 N A N A
Method Blank3 10.2 N A NA .
'Average of four, ranging from 0.1 to 1.2 pp 'Averape of four, ranginp from 0.1 to 0.8 p g "verage of three. ranpinp from 0.3 to 20.7 pp 'NA. Not Applicable
Table 5-6. Summary of Analytical Quality Control Results for Field Test #2 r
Total Detected Theoretical Percent Recovery Sample ID Y9 PQ %
Field Blank A 2.2 N A' N A
Field Blank B N D N A N A
Field Blank C 2.2 N A N A
Field Blank D N D N A N A
Field Spike 1 627 65 1 96.0
Field Spike 2 630 651 97.0
Method Spike' 628 651 96.0
Toluene Reapent Blank' ND N A N A
ACN Respent Blank' ND N A N A
Method Blank3 0.17 N A N A -
'Average of four. 'Averape of four. 'Average of eleven. ranging from 0.3 to 20.7 pp. 'NA, Not Applicable. nAverage of ten.
for Reld Test 13
Sample ID HDI MDI MI TDI I HDI MDI MI TDl
Total Detected Pg
f ie ld Blank A ND ND 0.6 ND
Field Blank B N D N D 1.1 ND
Field Blank C N D ND 0.7 ND
f ield Spike 1 163 170 2740 177
Method Spike' 158 163 2736 170
T duene Reagent Blank' ND ND 1 . ND
ACN Reagent Blank' ND N D 0.3 ND
Percent Qecoverf
Method Blank' N D N D 2.5 ND 1 NA N A N A N A
'Average of three, ranging from 0.1 to 1.2 pg. aAverege of three, ranging from 0.1 to 0.8 pg. a ~ v a r r g e of twelve, ranging from 0.3 to 20.7 p g 'NA = Not Applicable. 'Theoretical values HDI = 161
MDI = 165 M I = 2606 TO1 = 171 ND = Not Detected NA = Not Applicable
'Average of 6.
Section 6 Method Quality Control and
Method 301 Statistical Procedures
6.1 Quality Control
Tbe quality control procedures for fiekl d labratory activities not discussd in previous rections ue
d e s c n i below.
6.1. I Sampling O C Procedures
Tbe sampling QNQC program for this project incMed datn quality objectives, manual method sampling
perfonname criteria, field equipment calibrations, consistency of field spiking, sampling, md recovery procedures,
representative sampling, complete documentation of field data and abnormalities, md adequate field sample custody
procedures.
6 .1 .1 .1 Field Equipment Calibrations
Meter boxes, temperature gauges and probe nozzles were calrbroted rs d e s c n i in Section 4.2. Only S-Type
pitot tubes meding the requirements specified in tbe EPA's were used during the
samphg program. Pitot tubes were inspected aod documented as meeting EPA specifications prior to sampling.
6.1.1 .2 Sampling OperationIRecovery Procedures
To ensure consistency bemeen trainslruns, specified individuals conducted the manual sampling, and one
person was assigned to clean up, recover, and raassemble the gtasswue. This protocol serves to eliminate
propagation of multiple operator variance. All team members were fPmiliar with the procedures detailed in tbe test
plan for each fiekl test.
6.1.1 -3 Representative Sampling '
Each of the four dependent trains in tbe quad rsrembly was sampling essentially Lbe u m e stack gas.
Parameterr such as flow Rtes and flue gas -rphrre were cornpored to assess the variability in stnck gas conditions
for different test runs. (Flow rats were used for compprison plrp0ce.s only md u e provided in the test report for
reasons of confidentiality.) Any discrepancies were noted. Subsequent evaluation of uralyticpl results for
questionable runs was used to determioe whether results from inconsistent runs u e acceptable.
Communicrtion was Nintlinad betwan Rdian umpling personnel and l ine opmtors. Rdhn WM to be
ootified immediately of my iocidcnt which could b v e dTected tbe rtcsdy operation of tbe process. 7be field t&
M e r aIso me4 daily with appropriate produdion personnel to coordinate testing uauod possible product changss. It
was desirable to minimize production changes during the week-long test period. Such events were, bowever,
documented.
6.1.1.4 Documentation
Field data sbeets were complded lad thoroughly checked &r each test run. Velocity bead (AP)
determinations were required on only one of the four data rbests.
Any abnormalities observed during tbe test run or during rubsequent rewvery wpe recorded on the appropriate
field sheets. Any known variation from steady state (production) operations occurring during run periods was
documented. Documentation of pre- d post- test dibrations d inspections were maintained.
6.1.1.5 Sample Custody
Sample custody procedures for this program were based on EPA-recommended procedures. The custody
procedures emphasize careful documencation of sample collection d field d y t i c a l date d the use of chainef-
custody records for samples being transferred. These procedures are discussed below.
The sample recovery leader was responsible for ensuring that 111 samples taken were accounted for md that
proper custody ad documentation procedures were followed for the field sampling efforts. A master sample logbook
was maintained by the recovery task leader to provide a hardcopy of 111 sample collection activities. Manual flue gas
samphg data was also maintained by the recovery task l d e r .
Following sampie collection, each sample was given a unique alphanumeric sample identification code.
Sample labels and integrity seals were completed and affixed to the sample container. The sample v o w were
determined ad recorded d the liquid levels marked on each bottle. The sample identification code was recorded on
the sample label and in the sample logbook. Sample bottle lids were sealed on the outside with Teflonm tape to
prevent leakage. The samples were stored in a secure area until they were packed on ice for shipment to the
laboratory.
As tbe samples were packed for travel, chain-ofcustody fonns were completed for each shipment container.
Tbe cbain-ofcustody fonns and written instructions specifying the treatment of each sample was also eoclosed in the - sample shipment conminer. Shipping conlainers were labelled with arrows to clearly indicate tbe upright position of
sample bonles .
6.1.2 1 aboratory QA/QC Procedures
' he kbodory QA progrrm for thir projsd incMsd proper hadling, logging uxl tracking of incoming
umplcs, procedun vllidptions incMing HPLC w h m efficie.ncy, calibration curves, daily QC checks and rspl iae
. n r l y a , uxl wUsction a d o r d y r i r of umple, field rsd nrgent bknkr, rsd msthod cpika AS well AS field d
kboratory &pikes.
6.1.2.1 Sample Custody/Tracking
Upon rsceipc of umples in tbe laboratory, the cbainofcustody forms and umple bottle labels were recorded
to verify rsceipc of ump1e.s. Any didcnprocies or abnormalities (leakage, &.) were 4. Tbe samples were
stored at 4°C to minimize decomp~cition of derivatives.
6.1.2.2 Calibration Curve
A six-point calibration curve was prepared .nd d y z e d after initially seUing up the instrument a d whenever
tbe calibration check stardard did not fall within the f 5% W o w . Iftbe correlation caficient was less tban 0.995,
tbe cause was identified ant3 corrected.
6.1.2.3 Daily QC Checks
A check standard wac prepared d was used to check instrument response d the calibration curve. The
check standard was analyzed before d after all sample analyses d after eech sixth to eighth sample. Ifthe
calibration check response wns not within f 10% of the expected response, the cause was determined.
6.1.2.4 Blanks
Neat acetonitrile (system blank) was analyzed at least once per day to ensure that the system was not
contaminated. If a response was obtai3ed that is 10.1 of the level of tbe expected d y t e concentration, the source of
contamination was Icka~ed a d eliminated before analyzing samples.
6.1.2.5 Replicate Analyses
AU cnmples were analyzed in triplicate. Tbe triplicate raplyses should bave a coefficient of variation (I CV)
of 5 10% at concentrations greater than 1 pglmL d f 25% at concenlrations less than 1 pglrnL. lfthe replicate
d y r e s were outside of these tmits, the following items were checked:
% ~ J S u e integmed properly'; -
4 There is no interference from other components io the sample; ancl
The instrument is working properly.
6.1.2.6 Method Spikes
Ooe wtbod spike for every 10 cunplts wrs prspued. ?be r a c o v o h for method cpikom nwt be *20
percent. If tbe metbod spike recoveries w e n outride of this range, tbe clllse wrs identified d corrsdsd.
6.1.2.7 Field and Reagent Blanks
Field blanks were collected as dscnbed in S d o n 4.2.5. W e blrnks were collected d processed in the
came manner as collected samples.
Reagent btnks of recovery solvePts were also collsctbd d chipjmd to Rndian'r Morrisville, NC., laboratory.
Field Pod reagent blank analytical results serve u indicators of preparation d recovery contambation.
6.2 Method 301 Statistical Procedures
6.2.1 Calculation of the Amount of Isocyanate Collected
A least squares linear regression analysis of the dbro t ion data w u used to calculate a wrrelation coefficient,
slope, and intercept. Concentration was used as the independent or X-variable aDd response was used as the
depeodent or Y-variable.
Tbe wncentration of isocyanate (as the derivative) in the w n c m t ~ t e d samples was tbeo calculated u follows:
ComxNration = ( Sample Response - Intercept) Slope 6- I
Tbe total amount (pg) collected in r sample was then calculated by multiplying the concentration (pgJmL) times the
final volume (1 0 mL) of ACN used to redissolve the concentrated sample.
Amount TDI derisative = Concentration (pdmL) x F M Volume (10 mL)
Using TDI as an example, the equivalent amount of TDI required to generate this much derivntive w u crlculated by
multiplying the ratio of tbe molecular weights of TDI (174) Pod tbe TDI derivative (501) times tbe amount of TDI
derivative (determined by using equation 6-2).
6.2.2 Normefization of the Amount of Isocyanate Collected
In order to simplify the conrpuiron of tbe dy t ic * l rwultr of tbe four trrinr io wch QUA^ run for
mibsequd u l c u t t i o ~ s of birs rod pbcision, tbe test p h d d for tb c 0 U d o n of 30 ft' of sample in a c h train.
Due to operational vuiabilities inhered to u h train, -rate but slightly differing sample v o b rwuhsd.
Therefore, it wrs -suy to ~onrvlitc tbe Mouot of irocyuutc detected to a common sample volume. Tbe
following stepwise ulcuktiom wen used to omd dim tbe data. Tbe data from QUAD Run Number 1 field tsa #I
for TDI u e used u m example, wbere T& A rod B were dynrmicjly spiked and T h C rod D were unspiked.
Step 1 Normalize uspiked Train6 C rod D first impinger Mwdr (pg) to tbe r q l e vohurrc wllected (cubic
feet) in spiked Train A;
Train C 33'42 X 4409 pg = 4462 p g 33.02 A '
Train D 33A2 " X 4416 pg = 4467 pg 33.05 A 3
Step 2 . Average the normalized, unspiked train amounts from step 1 above. Assuming the collection efficiency
of all trains to be the same, this value would also be the unouot of TDI that would be collected by Train
A due to sampling tbe stack gas;
Step 3 Subtract the average value (slep 2) from lbe unwrrected amount (-led amount plus spike) for spiked
Train A, first impbeer, to pd recovered spike amount;
Step 4 Normalize the amount collected in spiked Train A, first impinger to 35.3 1 A' (35.3 1 A' which is equivalent to Im' was selected for.wnvenience only) using tbe sample volume for Train A and the value
- obtained in step 2;
~ P S Nornulize tbe .mard collected in tbe recoad irnpingor of spiked Trrin A to 35.31 ft' using the mmple volume for T n b A d the .mount dotomid by HPLC;
Step 6 ~un:tbevPluw determined fromdeps 3 , 4 d 5 lo gettbetotpl unount o f T D I f o d in~piked Train A, n o d to 35.3 1 ft';
7478 pg + 4718 pg + 67.5pg = 12264 pg 6-9
Steps 1-6 can be repeated for similar dcula&ions for spiked Train B. Tbe total rmount of isocyanate collscted in
each of tbe unspiked trains cen be determined by fint normalizing tbe mounts found in tbe two impingers of cacb
train to 35.31 ft' (i.e., 1x11') d tben summing tbe two vafues. Tbe raw d wrmalited data for dl analysis nsulls
are presented in Appedix A.
6.2.3 Precision of Spiked Samples
Precision of the spiked compounds was d c u l ~ l e d using tbe difference between the measured mnnalized
concentrations, 4, of tbe spiked compound for eocb spiked train. For this ulculation of precision to be valid, the
vPriances of tbe measured concentrations for cacb of the 6 runs must be the tame. If tbe variances were determined
to be uoequal, a review of the experimental data was performed to determine if here were any transcriplion errors or
m y irregular experimental procedures. If not, tbe 'unequal' vPriaace was flagged md pooled with the rest of tbe
variances according to Equation 6-10. Precision is reported as tbe st& deviation of tbe difference of spiked
c o m p d s , SD,, given by tbe following equation: .
wbere
n = tbe number of sampling mu used in the ulcula&ion (n = 6); md
d, = tbe difference of paired sampling train measurements.
Tbe standard deviation of tbe mean difference, SD,, was calculated as follows:
SD' - SD, = - J;;
where
n = tbe number of sampling mu (n = 6).
Tbe percent relative rtmd.rd deviation (or percent cosfficicot of vuirtioo) waa crlculrted u follows:
*re
SD, = srandard devhtion of the differc~ce of the spiked compouads; d
S, = mean of tbe spiked umpleb.
6.2.4 Precision of Unspiked Samples
Under tbe rssumplion of equal vui.nces for tbe 6 qrred mu, tbe precision of tbe 12 unspilrd samples was
ddermined by dculating the difference, d,, between tbe pin of tbe uspiked samples for each of tbe QUAD mas.
Tbe standard deviation of the unspiked values, SD,, wns dculated by:
SD, = - 4 :2
where
D = the number of sampling runs used in dculalion (n 6).
The percent relative sta& deviation (or percent coeficient of varh!ion) wns calculated as follows:
where
S, = tbe mean of the unspiked samples.
6.2.5 Bias
Bias was calculated using tbe dnh from tbe lnalysis of tbe 12 spiked fiekl samples d 12 unspikd field
uunples (assuming 6 QUAD Runs).
Bias of the spike cornpod is defrned as:
B = S , - h . 5 , - C S
- where
B = bias for the spiked compound;
S, = mean of the spiked samples;
= mean of the unspiked samples; d
CS = crlculat.ed vdue of tbe cpike compaund.
Tbe pooled ttPndnrd deviation, SD,, of tbe spiked md uatpiked trrins was ulcuktsd .r follow:
*re
SD, = standard deviation of the spiked Mrnple v-; a d
SD, = standard devirtion of the unspiked w l e v W .
Note that this equation overestimplec the standard devintion of tbe msan by a factor of the square root of 2.
Tbis is the approach recommended in Method 301, therefore it will be used for this validation procedure.
Tbe bias, B, calculated from Equation (6-15) was tested to determine if it is statistically different from zero. A
r-test was used to make this determination. Tbe t test compared the calculated r-statistic of the test data to the critical
r value for the specified degrees of freedom md level of sign5cance. For the test matrix in this plan, there are 6 data
points, which will be tested using a two-tailed r distribution af the 95% confidence level. Tbe t-statistic was
calculated .as shown in equation (6-17):
1 = (B - 0.01
SDM
where
B = bias;and
SDM = p l e d st& deviation of the mean (i.e., SD,I&).
This r test evaluates the hypothesis that the bias is greater than zero. Uibe calculated absolute value of tbe r-statistic
is greater than the two-tailed critical value for the specified degrees of freedom md level of significance, then there is
a 95 0X probability that the bias calculafed for tbe paired trnins is significant. U the calculated absolute value of the
r-statistic is less than the critical v h e for the fpecified degrees of freedom md level of sipficance, then tbe'averape
difference of the concentration between paired sampling trains is assumed to be zero md the measured cancentrations
cm be pooled for statistical tests. Tbe critical vdue of tbe f statistic for the two tailed tdistniution at 0.05 level of
significnnce (95 5C confidence level) with 5 degrees of freedom is 2.571. U the bias was found to be s ign i f id , then
a correction factor can be calculated to cornpewate for the bias. If the correction factor value falls between 0.7 and
1.3 the mctbod is still acceptable.
Section 7 References
Aklrich rod Mobay Material S a f q rod Dnh S b .
'Organic Chemistry', Morrison rod Boyd, AUyn rod Bacon, h., 1983, pp 844-845.
'Simulta~eous Ddennioation of Aromatic Isocy~nates rod Some Cucinogenic Amins in the Work Atmosphere by Reversed-Phase High-Performance Liquid Chromalography', E.H. Niemine, L.H. Saarioen and J.T. LPakso, Journal of Liquid Chromatogmphy ,6(3), 453469, 1983.
'Determination of Trace Atmospheric Isocyanate Concentrations by Reversed-Phase High-Performance Liquid Chromatography using 1-(2-Pyridyl) Piperazine Reagent', P.A. Goklberg, R.F. Walker, P.A. Ellwood, md H.L. Hardy, Journal of Chromatography, 212, 93-104, 1981.
Appedix B to Part 136. 'Definition d Procedure for the Ddermination of the Method Detection Limit.' Revision 1.11. FederalReeister. Volume 49, No. 209. October 26, 1984, 198-199.
'Recent Developments in the Sampling d Analysis of Isocyamdes in Air', V. Dhannar~jnn, R.D. Lingg, K.S. Booth d D.R. Hackathorn, American Society for Testing and Materials, Special Technical Publication 957, 1987.
Mitchell, William J., Midgett, M. Rodney, Means to Evalua!e Perfonnaoce of Stationary Source T a t Metbds , Environmental Science a d Technolopy, Vol. 10, Page 85, January 1976.
APPENDIX A
FIELD TEST #1 DATA
I
Figuro 1 First Impjnger, Unspiked Train
Second Impinger, Unspiked Train
Figure 3 First Impinger, Spiked Train
igum 4 Second Impinger, Spiked Train
Calibration Level 1 Concentration 1 Response Level 1 1 0.50 I 120958
I . .-
t Level 2 I 1.67 1 3
Level 2 1.67 3 Level 2 1.67 3 Level 3 2.50 5 Level3 I 2.50 1 6 - .
h v e l 3 2.50 6 Level 4 5.01 10 Level 4 5.01 10 bvel4 I 5.01 10
- - -
Level 5 6.68 14 W l S 6.68 14 L e d 6 8.35 17 bvd 6 8.35 17
L b v e l 7 I 12.52 1 2732793 Level 7 12.52 1 27mW
I
Level 6 I 8.35 L.vel7 12.52
1 7 W l 269 1 345
Results of Quad Run I , 2,4-Toluene Diisocyanate .
Date Sample AREA T D I Averag Analyze Name u8 U g
-
)3)25'/93' ARL- 12-2-0-2 13/25/93 ARL- 12-2-8-2 13/25/93 ARL- 12-2-0-2 13/25/93 ARL- 13-2-C- I 13/25/93 A RL- 13-2-C- I 13/25/93 ARL- 13-2-C- I 13/25/93 ARL- 14-2-C-2 )3/25/93 ARL- 14-2-C-2 13/25/93 ARL- 14-2-C-2
)3)2~./93i ARL- 16-2-D-
Results of Quad Run 2,2,4-Toluene Diisocyanate
Results of Quad Run 3, 2,4-Toluene Diisocyanate .
Date Sample AREA TDI ' Averag Analyze Name Ug WL!
Results of Quad Run 4,2,4-Toluene Diisocyanate
%CV Dry Standard Unspiked Recovered Amount Total % Total % Recovery Injection Meter Normalize Spike Normalized Normalize in 2nd from target '
(dscm) to Spike (ug) to 1 m3 (ug) 7827.81 ug
Results of Quad Run 5,2,4-Toluene Diisocyanate
Results of Quad Run 7, 2.4-Toluene Diisocyanate
Results of Quad Run 8, 2.4-To'luene Diisocyanate
APPENDIX B
FIELD TEST #2 DATA
Figurt 2 Second Impinger, Unspiked Train
r- I.. f . 5 8 c U* - 3~ - 5 5
igum 3 First Impinger, Spiked Train
Calibration Levell Concenuation 1 Response Lcvel 1 I 0.44 26437
1
Level 1 I 0.41 27156 Level 2 Level 2 Level 3 17U1 Level 3 1717
I I
Lcvel 4 I 5.2U 3360451
I
Level 5 I 7.8U 491406
Level 4 Level 5
I I
Level 6 1- 10.44 659424
1
5.24 33 1465 7.8a 48%1C
Level 6 Level 7
Level 10 I 24.84 l349358 Level 10 3.81 13546%
10.44 659504 l3.a 847342,
Level 7 Level 8 Level 8 Level 9 Level 9
U.04 832295 l5.6q 1012427 l5.6a 1018235 18.23 1162730 1 8 . q 117075&
1 Calibration Levell Concenlration 1 Response 1 L ~ v e l 1 I 0.44 Level 1 0.41 Level 2 Level 2
I 14 74601 13 72852,
L Level 3 Level 4 Level 4
Level 6 61939 Level 6 Level 7 Level 7 13.0
Level 5 Level 5
I Level 8 I
I 15.61 9330d
I
2.8 15 1968
i
7.84 44661 1 7.8U 437W
5.21 5.24
Level 10 20.81 1226891
300672 2 9 m
I I
Level 8 Level 9
15.64 918745 1 8 4 1078163
Pi-
Date Sample - Area MDI Averag %CV Analyze Name Count ug ug Injection
, 10'/27/93i ISO- 19-3-8- I ' 401 880' 685.51 689.57 0.5 1% 10/27/931 SO- 19-3-B- I 405144 691 -01 10/27/931 ISO- 19-3-B- I 405722 692.01
Dry Standard Unspiked Recovered Amount Total - % Total % Recovery ' Meter Normalize Spike Normalized Normalize in 2nd from target (dscm) to Spike (ue) to I m3 0%) 651 ug
Results for Quad Run 3. Methylene Diphenyl Diisocyanate
Results for Quad Run 4, Methylene Diphenyl Diisocyanate s
Analyze
- Area
Count MDI Averag %CV Dry Standard Unspiked Recovered
Ul3 ug Injection Meter Normalize Spike (dscm) to Spike (W)
Amount Normalized
to 1 m3
Total 1 % Total I % Recoverv
Results fat Quad Run 6, Methylene Diphenyl Diisocyanate
8n IS9 i a 8 m WOJJ 1S~ar-x%
r
puz u! ~ r l o l %
(an) a q u r u ~ o ~
la04
0" I 01 pazgerulo~
iunoru y
(an) aq!ds
p a l a ~ o 3 a ~
aq!ds 08 az ! l e lu~o~ pay !dsun
( ~ ~ D s P ) l a j a ~
p~epue~g L J ~ u o ! p a ~ u ~
~ 3 % Bn
Be'aa y an
law luno3 e a ~ v
aweN aldwes
nLleu y a 8 ~
Analyze %CV 1 Drv Standard1 Uns~iked I Recovered Averag
ug
Results for Quad Run 10, Methylene Diphenyl Diisocyanate
APPENDIX C
FIELD TEST #3 DATA
First Impinger, Unspiked Train
Figure 2 Second Impinger, Unspiked Train
Figure 3 First Impinger, Spiked Train
Second Impinger, Spiked Train
f Calibration Level l Concentralion 1
I level 1 I I
I 1.W 50514 t level 2
I
I 2.01ll 1012091
t level 3 I
I 4.0~4 203nd
Level 2 level 3
I J
2.014 101826 4.0a 201489
t Level 6 I I
I 20.1ll 985864
' ~ c v e ~ 5 Level 5
I Level 6 I I
I ~3.131 98gad
I J
15.081 74 1034 . 15.M 736319
Level 7 level 8 Level 8 Level 9
I I I
Level 6 I 16.14
Calibration Level Concentration ( Response h v e l 1 0329q 24244 Level 1 0329a
Level 2 Level 3 Level 3 Level 4 55527 Len1 4 564 1 1
t Level 6 I
I 16.46 1 0 9 5 6 4
Area ( MI 1 Average Count ug Ug
%CV I Dry 1 Unspiked 1 Recovered I Amount I Total I % Total Injection Meter Normalize Spike NormalizedNonnalize in2nd
(dscm) tospike (ue) to I m3 (UB) (ug)
Results for Quad Run 3, Methyl Isocyanate
from ta~get 10060 UB
Analyze TP"- Area
Count Injection Meter (dscm)
Unspiked 1 Recovered 1 Amount 1 Total ( % Total % Recovery from target
10060 ug
Results for Quad Run 4, Methyl Isocyanate
Results for Quad Run 6, Methyl Isocyanate
I Date I Sample Count ug
Results for Quad Run 8, Methyl Isocyanate
Dry Standard Meter (dscm)
Uns~iked l Recovered Amount I Total I % Total 1 % Recovery 1 Normalized Normalize in 2nd from target
to 1 m3 I (us) I I l m ug I
Date Sample Area MI Average %CV Analyze 1 Count ug ug Injection
Dry Standard Meter (dscm)
Unsoiked l Recovered 1 Amount I Total I % Total I % Recovery 1 No&a~i=e 1 Spike I to Spike ( w ) to I m3 I
Results for Quad Run 9, Methyl Isocyanate
Results for Quad Run I I , Methyl Isocyanate
Results for Quad Run 12, Methyl Isocyanate
Sample Analyze Count
ARL- 14-2C2 1524 0.2 ARL- IS-2DI 304294 235.0 ARL- IS-2DI 307565 237.5 ARL-15-2DI 306133 236.4 ARL-16-2D2 3200 0.7 ARL-16-2D2 3854 0.9 A RL- 16-2D2 3238 0.8
Results from Quad Run 2, Hexamethylene Diisocyanate
Date Sample Area HDI Averag %CV Dry Standard Unspiked Recovered Amount - Total % Total % Recovery Analyze Name Count ug ug Injection Meter Normalize Spike Normalized Normalize in 2nd from target
(dscm) to Spike (ug) to I m3 ( ~ 8 ) 161.4 ug
'06/28/94 06/28/94 06/28/94 06/28/94 06/28/94 - 06/29/94 06/29/94 06/29/94 06/29/94 06/29/94 06/29/94 06/29/94 06/29/94 06/29/94 06/29/94 06/29/94 06/29/94
Results from Quad Run 5, Hexamethylene Diisocyanate
ARL-35-5B I ARL-35-5B I ARL-36-5B2 A R L- 36-5 B2 ARL-36-5B2 ARL-37-5CI ARL-37-5Cl ARL-37-5C1 A RL-38-5C2 ARL-38-5C2 A RL-38- SC2 ARL-39-5DI A RL-39-5DI A RL-39-501 A RL-40-SD2 ARL-40-SD2 ARL-40-5D2
238248 238594
464 839 446
32427 32798 32464
428 445 476
30468 32039 3246 1
610 0
780
183.8 184.1
0.0 0.0 0.0
24.5 24.8. 24.5 0.0 0.0 0.0
23.0 24.2 24.5 0.0 0.0 0.0
0.00
24.59
0.00
23.89
0.00
NA
0.65%
NA
3.4 1 %
NA
0.00
23.72
0.00
21.94
0.00
I
1 .036453234
1.088499599
23.72
2 1.94
0.0%
0.0%
I
Results from Quad Run 6, Hexamethylene Diisocyanate
. Date Sample Area HDI Averag
Analyze 1 Count ug Ug %CV [Dry standard
Injection Meter I - (dscm)
Vnsmked l Recovered I Amount I Total I % Total 1 % Recoverv 1 ~ormal ize Spike Normalized Normalize in 2nd from target t o s p i k e 1 (ug) 1 t o l d I (ug) I , I 1 6 1 1 u g I
(uu)
Results from Quad Run 7, Hexamethylene Diisocyanate
Sample Unspi ked Recovered Amount Total % Total Normalize Spike Normalized Normalize in 2nd to Spike to I m3
- -
Results from Quad Run 8, Hexamethylene Diisocyanate
Results from Quad Run 10, Hexamethylene Diisocyanate
Results from Quad Run 12, Hexamethylenc Diisocyanate
Results o f Quad I, 2.4-Toluene Diisocyanate
Date Analyze
Sample Name
'06/07/94 06/07/94
06/07/94 06/07/94 06/07/94
06/07/94 06/07/94
06/07/94 06/07/94 06/08/94 06/08/94 06/08/94 06/08/94 06/08/94 06/08/94 06/08/94 06/08/94 06/08/94
06/08/94
06/08/94 06/08/94
Area Count
ARL-02- 1 A2 ARL-02- I A2
ARL-02-IA2 ARL-03- I BI ARL-03- I BI
ARL-03- I B I '
ARL-04- 1 82
ARL-04- I BZ A RL-04- I B2
ARL-05- IC I A RL-05- I C I ARL-05- I C I ARL-06- IC2 ARL-06- 1C2 ARL-06- I C2 ARL-O7- ID I ARL-07- I D l
ARL-07- I Dl
ARL-08- ID2 ARL-08- I D2 ARL-08- I D2
TDI
ug
1250 1810
2041 417128 422140
425883
O
784
448 7402 7198 7434 652 612
0
7453 7612
O O 0
Averag
ug
0.0 0.0
0.0 169.6 17 1.6 173.1
0.0
0.0 0.0 1.9 1.8 1.9 0.0 0.0 0.0
1.9 2.0
0.0
0.0 0.0
%CV Injection
0.00
171.44
0.00
1.87
0.00
- -3 .57
0.00
Dry Standard Meter (dscm)
'NA
1.05%
NA
2.8 1 %
NA
NA
Unspiked Normalize
to Spike
(ug)
1.103535844
1 .09880693 1
Recovered Spike
(ul3)
2.18
Amount Normalized
to I m3
169.26
I
Total Normalize
( ~ 8 )
0.00
% Total in 2nd
I
4 1 . 8 0 4 6 1 . 1 4 O 9 4 2 3 9 9 - 2 , 2 3 ~ ~ ~
% Recovery from target
171.2 ug
0.00
0.00
0.0%
0.0%
.
1.98
0.00
1.70
171.24
1.70
g Analyze %CV 1 Dry Standard 1 Unspiked 1. Recovered Amount I Total 1 % Total 1 % Recovery Normalized Normalize in 2nd from target
to I m3 (W) 171.2 ug
Results of Quad 2,2,4-Toluene Diisocyanate
Results of Quad 4, 2,4-Toluene Diisocyanate
Sample
Results of Quad 5, 2,4-Toluene Diisocyanate
(dscm) to Spike
(ug) - 2.33
Recovered 1 Amount 1 Total 1 % Total I % Recovery Spike Normalized Normalize in 2nd from target
(ug) to 1 m3 b e ) 171.2 ug
Results of Quad 6, 2,4-Toluene Diisocyanate
Results of Quad 7, 2.4-Toluene Diisocyanate
Results of Quad 8.2.4-Toluene Diisocyanate
Results of Quad 10.2.4-Toluene Diisacyanate
* Date - Sample Area TDI Average %CV Analyzed 1 Count ug ug Injection
Results o f Quad I I, 2.4-Toluene Diisacyanate
Results of Quad 12, 2.4-Toluene Diisacyanate
Analyze
Results of Quad Run 3, Methylene Dipht nyl Diisocyanate
yr i f - < 0
Results of Quad Run 6, Methylene Diphenyl Diisocyanate
Results of Quad Run 7, Methylent Diphenyl Diisocyanate
Analyze I=-
I i I 0.00 INA 0.00
I I I
Rwulb of Quad Run 8, Methylent Diphenyl Diisacyanate
Analyze r
Results of Quad Run 9, Methylene Diphenyl Diisocyanate
Analyze r r
Results of Quad Run 10, Methylene Diphenyl Diisocyanate
,07/30/94-ARL-88-IID2 - 1101- 0.0 0.00 INA 0.00 07/30/94 ARL-88- I 1 D2 22 13 0 .0 ,07/30/94 ARL-88- 1 1 D2 1310 0.0 '
Results of Quad Run 1 1 , Methylene Diphenyl Diisocyanate
Results of Quad Run 12, Methylene Diphenyl Diisocyanate
APPENDIX D
SAMPLING METHOD FOR ISOCYANATES
METHOD XXXl
METHOD XXXI - SAMPLING METHOD FOR ISOCYANATES
NOTE: This method is not inclusive with respect to specifications (e.g., equipment and supplies) and sampling procedures essential to its perfonnance. Some material is incorporated by reference fmm other EPA methods. Therefore, to obtain reliable results, persons using this method should have a thorough knowledge of at least the following additional test methods found in 40 CFR Part 60: Method 1, Method 2, Method 3, and Method 5.
1.0 Scope and Application.
1.1 This method is applicable to the collection of isocyanate compounds from the emissions associated with manufacturing processes. The following is a list of the isocyanatcs and the manufacturing process at which the method has been evaluated:
Detection Limits Manufacturing Compound Name CAS No. (nglm')' Process
2,4-Toluene Diisocyanate 584-84-9 106 Flexible Foam 0-w Production
I ,6-Hexamethylene 822-06-0 396 Paint Spray Booth Diisocyanate (HDI)
Methylene Diphenyl 10 1-68-8 112 Pressed Board Diisocyanate (MD1) Production
Methyl Isocyanate (MI) 624-83-9 228 Not used in production
s Estimated detection limits are based on a sample volume of 1 d and a 10 mL sample extraction volume.
2.0 Summary of hlethod.
2.1 Gaseous and/or aerosolized isocyana~es are withdrawn from an emission source at an isokinetic sampling rate and are collected in a multicomponent sampling train. The primary components of the train include a heated probe, three impingers containing the derivatizing reagent in toluene, an empry impinger, an impinger containing charcoal and an impinger containing silica,gel.
2.2 The collected samples are analyzed by high performance liquid chromatography (HPLC) as described in Method XXX2.
3.0 Defi~itions. Not Applicable.
4.1 The greatest potential for interference comes from an impurity in the derivatizing reagent, I-(2-pyridy1)piperazine (1,2-PP). This compound may interfere with the resolution of MI from the peak attributed to unreacted I ,2-PP.
4.2 Other interferences that could result in positive or negative bias are (I) alcohols that could compete with the 1.2-PP for reaction with an isocyanate and (2) other compounds that may coelute with one or more of the derivatized isocyanates. -
5.1 The toxicity of each reagent has been precisely defined. Each isocyanate can produce dangerous levels of hydrogen cyanide (HCN). The exposure to these chemicals must be reduced to the lowest possible level by whatever means available. The laboratory is responsible for maintaining a current awareness file of Occupational
Revision 2 August 1995
METMOD XXXl
Safety and Health Administration (OSHA) regulations regarding safe handling of the chemicals specified in this method. A reference file of material safety data sheets should also be made available to all personnel involved in the chemical analysis. Additional references to laboratory safety are available.
6..O Equipment and Supplies.
6.1 Sample Collection. The following items are required for sample collection: 6.1 .I A schematic of the sampling train used in this method is shown in Figure XXXI-I. This sampling
train configuration is adapted from EPA Method 5 procedures, and, as such, the majority of the required equipment is identical to that used in EPA Method 5 determinations. The only new component required is a condenser coil.
6.1.2 Construction details for the basic train components are given in APTD-0581 (see Martin, 1971, in Section 16.0, References); commercial models of this equipment are also available. Additionally, the following subsections list changes to APTD-0581 and identify allowable train configuration modifications.
6.1.3 Basic operating and maintenance procedures for the sampling train are described in APTD-0576 (see Rom, 1972, in Section 16.0, References). As correct usage is imponant in obtaining valid results, all users should refer to APJD-0576 and adopt the operating and maintenance procedures outlined therein unless otherwise specified. The sampling train consists of the components detailed below.
6.1.3.1 Probe Nozzle. Glass with sharp, tapered (3P angle) leading edge. The taper shall be on the outside to preserve a constant internal diameter. The nozzle shall be buttonhook or elbow design. A range of nozzle sizes suitable for isokinetic sampling should be available in increments of0.16 cm ( 1 4 6 in.), e.g., 0.32-1.27 cm (1/8-1/2 in.), or larger if higher volume sampling trains are used. Each nozzle shall be calibrated according to the procedures outlined in Paragraph 10.1.
6.1.3.2 Probe liner. Borosilicate or quartz-glass tubing with a heating system capable of maintaining a probe gas temperature of 120 rt 14 'C (248 i 25 'F) at the exit end during sampling. (The tester may opt to operate the equipment at a temperature lower than that specified.) Because the actual temperature at the outlet of the probe is not usually monitored during sampling. probes constructed according to APTD-0581 and utilizing the calibration curves of APTD-0576 (or calibrated according to the procedure outlined in APTD-0576) are considered acceptable. Either borosilicate or quartz glass probe liners may be used for stack temperatures up to about 480'C (900 'F). Quartz glass liners shall be used for temperatures between 480 and 900'C (900 and 1650 'F). (The softening temperature for borosilicate is 810 'C (1508 'F), and for quanz glass 1500 'C (2732 'F).) Water-cooling of the stainless steel sheath ~ v i l l be necessary at temperatures approaching and exceeding 500'C.
6.1.3.3 Pitot tube. Type S, as described in Section 2.1 of promulgated EPA Method 2 (Section 6.1 of Reformatted Draft EPA Method 2), or other appropriate devices (see Vollaro, 1976 in Section 16.0, References). The pitot tube shall be attached to the probe to allow constant monitoring of the stack-gas velocity. The impact (high-pressure) opening plane of the pitot tube shall be even with or above the nozzle entry plane (see EPA Method 2. Figure 2-6b) during sampling. The T}pe S pitot tube assembly shall have a known coefficient, determined as outlined in Seclion 4.0 of promulgated EPA Method 2 (Section 10.0 of Reformatted Draft EPA Method 2).
6.1.3.4 Differential Pressure Gauge. Inclined manometer or equivalent device as described in Section 2.2 of promulgated EPA Method 2 (Section 10.0 of Reformatted Draft EPA Method 2). One manometer shall be used for velocity-head (delta P) readings and the other for orifice differential pressure (delta H) readings.
6.1.3.5 lmpinger Train. Six 500 mL impingers are connected in series with leak-free ground-glass joints following immediately afier the heated probe. The first impinger shall be ofthe Greenburg-Smith design with the standard tip. The remaining five impingers shall be of the modified Greenburg-Smith design, modified by replacing the tip irith a 1.3-cm (10-in.) 1.D. glass tube'extending about 1.3 cm ( ID in.) from the bottom of the outer cylinder. The first, second and third impingers shall contain known quantities of the derivatizing reagent in toluene with the first impinger containing 300 mL and 200 mL in the second and third. The fourth impinger remains empty. The . fifth impinger is filled with a known amount (215 full) of activated charcoal and the sixth with a known amount of desiccant. A water jacketed condenser is placed between the outlet of the first impinger and the inlet to the second impinger to minimize the evaporation of toluene from the first impinger. -
Revision 2 August 1995
METHOD XXXl
Figure XXX 1 - 1 . Isocyanate Sampling Train
Revision 2 August 1995
METHOD XXXl
6.1.3.6 Metering System. The necessary components are a vacuum gauge, leak-free pump, temperature sensors capable of measuring temperature to within 3 % (5.4 OF), dry-gas meter capable of measuring volume to within I%, and related equipment, as shown in Figure XXXI-I. At a minimum, the pump should be capable of 4 cubic feet per minute (cfm) free flow, and the dry-gas meter should have a recording capacity of 0-999.9 cubic feet (cu ft) with a resolution of 0.005 cu ft. Other metering systems capable of maintaining sampling rates within 10% of isokineticity and of determining sample volumes to within 2% may be used. The metering system must be used in conjunction with a pitot tube to enable checks of isokinetic sampling rates. Sampling trains using metering systems designed for flow rates higher than those described in APTD-0581 and APTD-0576 may be used, provided that the specifications of this method are met.
6.1.3.7 Barometer. Mercury, aneroid, o r other barometer capable of measuring atmospheric pressure to within 2.5 mm Hg (0.1 in. Hg). In many cases the barometric reading may be obtained from a nearby National Weather Service station, in which case the station value (which is the absolute barometric pressure) is requested and an adjustment for elevation differences between the weather station and sampling point is applied at a rate of minus 2.5 mm Hg (0.1 in. Hg) per 30-m (100 ft) elevation increase (vice versa for elevation decrease).
6.1.3.8 Gas density determination equipment. Temperature sensor and pressure gauge (as described in Sections 2.3 and 2.4 of promulgated EPA Method 2 (Sections 6.3 and 6.4 of Reformatted Drafi EPA Method 2)), and gas analyzer, if necessary (as described in EPA Method 3). The temperature sensor ideally should be permanently attached to the pitot tube or sampling probe in a fixed configuration such that the tip of the sensor extends beyond the leading edge of the probe sheath and does not touch any metal. ~lternatively, the sensor may be attached just prior to use in the field. Note, however, that if the temperature sensor is attached in the field, the sensor must be placed in an interference-free arrangement with respect to the Type S pitot tube openings (see promulgated EPA Method 2, Figure 2-7 (Reformatted Drafi EPA Method 2, Figure 2-4)). As a second alternative, if a difference of no more than 1% in the average velocity measurement is to be introduced, the temperature sensor need not be attached to the probe or pitot tube.
6.1.3.9 Calibration/Field-Preparation Record. A permanently bound laboratory notebook, in which duplicate copies of data may be made as they are being recorded, is required for documenting and recording calibrations and preparation procedures (i.e., silica gel tare weights, quality assurance/quality control check results, dry-gas meter, and thermocouple calibrations, etc.). The duplicate copies should be detachable and should be stored separately in the test program archives.
6.2 Sample Recovery. The following items are required for sample recovery: 6.2.1 Probe Liner. Probe and nozzle brushes; Teflon@ bristle brushes with stainless steel wire or Teflon@
handles are required. The probe brush shall have extensions constructed of stainless steel, Teflon@, or inen material at least as long as the probe. The brushes shall be properly sized and shaped to brush out the probe liner and the probe nozzle.
6.2.2 Wash Bottles. Three. Teflon53 or glass wash bottles are recommended; polyethylene wash bottles should not be used because organic contaminants may be extracted by exposure to organic solvents used for sample recovery.
6.2.3 Glass Sample Storage Containers. Chemically resistant, borosilicate amber glass bottles, 500-mL or 1,000-mL. Bottles should be tinted to prevent action of light on sample. Screw-cap liners shall be either Teflon03 or constructed to be leak-free and resistant to, chemical attack by organic recovery solvents. Narrow-mouth glass bottles have been found to exhibit less tendency toward leakage.
6.2.4 Graduated Cjflinder andtor Balances. To measure impinger contents to the nearest 1 mL or 1 g. Graduated cylinders shall have subdivisions not >2 mL. Laboratory balances capable of weighing to i0 .5 g or better are required.
6.2.5 Plastic Storage Containers. Screw-cap polypropylene or polyethylene containers to store silica gel and charcoal.
6.2.6 Funnel and Rubber Policeman. To aid in transfer of silica gel or charcoal to container (not necessary if silica gel is weighed in field).
6.2.7 Funnels. Glass, to aid in sample recovery. 6 3 Crushed Ice. Quantities ranging from 10-50 lb may be necessary during a sampling run, depending on
ambient air temperature. 6.4 Stopcock Grease. The use of silicone grease is not permitted. Silicone grease usage is not necessary if
screw-on connectors and Teflon93 sleeves or ground-glass joints are used.
Revision 2 August 1995
METHOD XXXl
7.0 Reagents and Stanahis.
7.1 Sample Collection Reagents. 7.1.1 Charcoal. Activated, 6- 16 mesh. Used to absorb toluene vapors to prevent them from entering the
metering device. Use once with each train and discard. 7.1.2 Silica Gel. Indicating type, 6-16 mesh. If previously used, dry at 175'C (350 'F) for 2 hours before
using. New silica gel may be used as received. Alternatively, other types of desiccants (equivalent or better) may be used, subject to the approval of the Administrator.
7.1.3 lmpinger Solution. The impinga solution is prepared in the laboratory by mixing a known amount of 1 - ( 2 ~ ~ ~ r i d ~ l ) piperazine (purity 99.5+ %) in toluene (HPLC grade or equivalent). The actual concentration of 12-PP should be approximately four times the amount needed to ensure that the capacity of the derivatizing solution is not exceeded. This amount shall be calculated from the stoichiometric relationship between 1.2-PP and the isocyanate of interest and preliminary information about the concentration of the isocyanate in the stack emissions. A concentmion of 130 ug/mL of 1.2-PP in toluene can be used as a reference point. This solution should be prepared ill the laboratory, stored in a refrigerated area away from light, and used within ten days of preparation.
7 2 Sample Recovery Reagents. 72.1 Toluene. Distilled-in-glass grade is required for sample recovery and cleanup (see NOTE to 7.2.2
below). 7.22 Acetonitrile. Distilled-in-glass grade is required for sample recovery and cleanup.
NOTE: Organic solvents from metal containers may have a high residue blank and should not be used. Sometimes suppliers transfer solvents from metal to glass bonles; thus blanks shall be run prior to field use and only solvents with a low blank value (<0.001%) shall be used.
8.0 Sample Collection, Preservation, Storage and Tramport.
8.1 Because of the complexity of this method, field personnel should be trained in and experienced with the test procedures in order to ottain reliable results.
8.2 Preliminary Field Determinations. 8.2.1 Select the sampling site and the minimum number of sampling points according to EPA Method 1 or
as specified by the Administrator. Determine the stack pressure, temperature, and range of velocity heads using EPA Method 2. It is recommended that a leak-check of the pitot lines (see promulgated EPA Method 2, Section 3.1 (Reformatted Draft ~ ~ ~ ' ~ e t h o d 2, Section 8.1)) be performed. Determine the stack gas moisture content using EPA Approximation Method 4 or its alternatives to establish estimates of isokinetic sampling-rate settings. Determine the stack-gas dry molecular weight, as described in promulgated EPA Method 2, Section 3.6 (Reformatted Draft EPA Method 2, Section 8.6). If integrated EPA Method 3 sampling is used for molecular weight determination, the integrated bag sample shall be taken simultaneously with, and for the same total length of time as, the sample run.
8.2.2 Select a noule size based on the range of velocity heads so that it is not necessary to change the noule size in order to maintain isokinetic sampling rates. During the run, do not change the noule. Ensure that the proper differential pressure gauge is chosen for the range of velocity heads encountered (see Section 2.2 of promulgated EPA Method 2 (Section 8.2 of Reformaned Drafi EPA Method 2)).
8.2.3 Select a suitable probe liner and probe length so that all traverse points can be sampled. For large stacks, to reduce the length of the probe, consider sampling from opposite sides of the stack.
8.2.4 A typical sample volume to be collected is I dscm (35.31 dscf). The sample volume can be adjusted as necessitated by analy~ical detection limit constmints andlor estimated stack concenrrations. A maximum limit should be determined to avoid exceeding the capacity of the reagent.
8.2.5 Determine the total length of sampling time needed to obtain the identified minimum volume by comparing the anticipated average sampling rate with the volume requirement. Allocate the same time to a11 traverse points defined by EPA Method 1. To avoid timekeeping errors, the length of time sampled at each traverse point should be an integer or an integer plus one-half min.
8.2.6 In some circumstances (e.g., batch cycles) it may be necessary to sample for shorter times at the traverse points and to obtain smaller gas-sample volumes. In these cases, the Administrator's approval must first be obtained.
Revision 2 August 1995
METHOD XXXl
8.3 Preparation of Sampling Train. 8.3.1 During preparation and assembly of the sampling train, keep all openings where contamination can
occur covered with Teflon03 film or aluminum foil until just prior to assembly or until sampling is about to begin. 8.3.2 Place 300 mL of the impinger absorbing solution in the f im impinger and 200 mL each in the second
and third impingers. The fourth impinger shall remain empty. The fifth and sixth impingers shall have 400 g of preweighed charcoal and 200-300 g of silica gel, respectively.
8.3.3 When glass probe liners are used, install the selected nozzle using a Viton@-A O-ring when stack temperatures are <260 'C (500 'F) and a woven glass-fiber gasket when temperatures are higher. See APTD-0576 mom, 1972) for details. Other connecting systems utilizing Teflon@ femles may be used. Mark the probe with heat-resistant tape or by some other method to denote the proper distance into the stack or duct for each sampling point.
8.3.4 Set up the train as shown in Figure XXXI-I. During assembly, do not use any silicone grease on ground-glass joints. Connect all temperature sensors to an appropriate potentiometerldisplay unit. Check all temperature sensors at ambient temperature.
8.3.5 Place crushed ice around the impingers. 8.3.6 Turn on the condenser coil coolant recirculating pump and begin monitoring the gas entry
temperature. Ensure proper gas entry temperature before proceeding and again before any sampling is initiated. It is imponant that the gas entry temperature not exceed 50% (122 'F), thus minimizing the loss oftoluene from the first impinger.
8.3.7 Turn on and set the probe heating systems at the desired operating temperatures. Allow time for the temperature to stabilize.
8.4 Leak-Check Procedures. 8.4.1 Pre-test leak-check. 8.4.1 .I Because the number of additional intercomponent connections in the train (over the EPA Method 5
Train) increases the possibility of leakage, a pre-test leak-check is required. 8.4.1.2 Afier the sampling train has been assembled, turn on and set the probe heating systems at the
desired operating temperatures. Allow time for the temperatures to stabilize. If a Viton@ A O-ring or other leak-free connection is used in assembling the probe nozzle to the probe liner, leak-check the train at the sampling site by plugging the nozzle and pulling a 381-mm HE, (15-in. Hg) vacuum. Leakage rates in excess of 4% of the average sampling rate or >0.00057 m7/min (0.020 cfm), whichever is less, are unacceptable.
NOTE: A lower vacuum may be used, provided that it is not exceeded during the test.
8.4.1.3 The following leak-check instructions for the sampling train described in APTD-0576 and APTD-058 1 may be helpful. Start the pump with the fine-adjust valve fully open and the coarse-adjust valve completely closed. Panially open the coarse-adjust valve and slowly close the fine-adjust valve until the desired vacuum is reached. Do mt reverse direction of the .fine-adjust valve; this will cause impinger contents to back up in the train. If the desired vacuum is exceeded, either leak-check at this higher vacuum or end the leak-check, as shown below, and stan over.
8.4.1.4 When the leak-check is completed, first slowly remove the plug from the inlet to the probe. When the vacuum drops to 127 mm (5 in.) Hg or less, immediately close the coarse-adjust valve. Switch off the pumping system and reopen the fine-adjust valve. Do not reopen the fine-adjust valve until the coarse-adjust valve has been closed. This prevents the reagent in the impingers from being forced backward into the probe and silica gel from being entrained backward into the fifth impinger.
8.4.2 Leak-Checks During Sampling Run. 8.4.2.1 If, during the sampling run, a component change becomes necessary, a leak-check shall be
conducted immediately after the interruption of sampling and before the change ismade. The leak-check shall be done according to the procedure outlined in Paragraph 8.4.1, except that it shall be done at a vacuum greater than or equal to the maximilm value recorded up to that point in the test. If the leakage rate is found to be no greater than 0.00057 m3/min (0.020 cfm) or 4% of the average sampling rate (whichever is less), the results are acceptable, and no correction will nced to be applied to the total volume of dry gas metered. If a higher leakage rate is obtained, the tester shall void the sampling run.
NOTE: Any "correction" of the sample volume by calculation reduces the integrity of the pollutant concentration data generated and must be avoided.
Revision 2 August 1995
8.4.22 Immediately after a component change, and before sampling is reinitiated, a leak-check similar to a pre-test leak-check must also be conducted.
8.4.3 Post-Test Leak-Check. 8.4.3.1 A leak-check of the sampling train is mandatory at the conclusion of each sampling run. The
leak-check shall be performed with the same procedures as those with the pre-test leak-check, except that it shall be conducted at a vacuum greater than or equal to the maximum value reached during the sampling run. lf the leakage rate is found to be no greater than 0.00057 d/min (0.020 cfm) or 4% of the average sampling rate (whichever is less), the results are acceptable, and no correction need be applied to the total volume of dry gas metered. If, however, a higher leakage rate is obtained, the tester shall either record the leakage rate, correct the sample volume (as shown in Section 12.0 of this method), and consider the data obtained of questionable reliability, or void the sampling run.
8.5 Sampling-Train Operation. 8.5.1 During the sampling run, maintain an isokinetic sampling rate to within 10% of true isokinetic, unless
otherwise specified by the Administrator. 8.5.2 For each run, record the data required on a data sheet such as the one shown in Figure
XXX1-2. Be sure to record the initial dry-gas meter reading. Record the dry-gas meter readings at the beginning and end of each sampling time increment, when changes in flow rates are made before and after each leak-check, and when sampling is halted. Take other readings indicated by Figure XXX1-2 at least once at each sample point during each time increment and additional readings when significant changes (20% variation in velocity-head readings) necessitate additional adjustments in flow rate. Level and zero the manometer. Because the manometer level and zero may drift due to vibrations and temperature changes, make periodic checks during the traverse.
8.5.3 Clean the stack access ports prior to the test run to eliminate the chance of collecting deposited material. To begin sampling, verify that the probe heating system is at the specified temperature, remove the noule cap, and verify that the pitot tube and probe are properly positioned. Position the noule at the first traverse point, with the tip pointing directly into the gas stream. Immediately start the pump and adjust the flow to isokinetic conditions. Nomographs, which aid in the rapid adjustment of the isokinetic sampling rate without excessive computations, are available. These nomographs are designed for use when the Type S pitot-tube coefficient is 0.84 * 0.02 and the stack-gas equivalent density (dry molecular weight) is equal to 29 4. APTD-0576 details the procedure for using the nomographs. If the stack-gas molecular weight and the pitot-tube coefficient are outside the above ranges, do not use the nomographs unless appropriate steps (Shigehara, 1974, in Section 16.0, References) are taken to compensate for the deviations.
8.5.4 When the stack is under significant negative pressure (equivalent to the height of the impinger stem), take care to close the coarse-adjust valve before inserting the probe into the stack, to prevent the impinger solutions From backing into the probe. If necessary, the pump may be turned on with the coarse-adjust.valve closed.
8.5.5 When the probe is in position, block off the openings around the probe and stack access port to prevent unrepresentative dilution of the gas stream.
8.5.6 Traverse the stack cross section, as required by EPA Method I or as specified by the Administrator, being careful not to bump the probe noule into the stack walls when sampling near the walls or when removing or inserting the probe through the access port, in order to minimize the chance of extracting deposited material.
8.5.7 During the test run, make periodic adjustments to keep the temperature of the condenser at the proper levels; add more ice and, if necessary, salt to maintain the temperature. Also, periodically check the level and zero of the manometer.
8.5.8 A single train shall be used for the entire sample run, except in cases where simultaneous sampling is required in two or more separate ducts or at two or more different locations within the same duct, or in cases where equipment failure necessitates a change of trains. In all other situations, the use of two or more trains will be subject to the approval of the Adminisnator.
8.5.9 At the end of the sample run, close the coane-adjust valve, remove the probe and noule from the stack, turn off the pump, record the final dry-gasmeter reading, and conduct a post-test leak-check. Also,
-
Revision 2 August 1995
FIELD DATA
1 Ambicnl Tcmpcrature ('F1 1 I Probe Heater Setting ('F) I I OZC02 Method I I
Samphq L d h
Sampb Tm
Run Number
Opcrala
Melec Box Number
Ma AH@
Yd
K Facla
Baranelrk Pressure (In)
Stalk Pressure (in H20)
- - -
F l k a Number
Assumed Mdsture (%)
0 2
coz (%)
e
Read and Recad a l Data Every Mlnutu bl.armofbuct
Healer Box Setting (OF)
lnit~al Leak Check
Travsre Potnt .
Nunbcr
--
Carmcnts:
-- - - - - -
Mdsture Cdkcled (g)
Flnal Leak Check
Smpl lng T h (mln)
Oar M d n Reading Vm (n')
Clock T h ( U h r )
vekcny Herd AP
(In ~ 2 0 )
O r i k s Pressure
Differenllal AH (In H20)
Flue Gar Temperalure
('F)
Probe Temperatun
('F)
FMer Tanperdure
(OF)
Dry Gar M e l a Trmpsrature
mplnocr- Ex# ('F)
Ink( rf)
V m (h. Hg)
0ulM ('F)
METHOD XXXl
leak-check the pitot lines as described in EPA Method 2. The lines must pass this leak-check in order to validate the velocity-head data.
85.10 Calculate percent isokineticity (see Section 12.8) to determine whether the run was valid or another test run should be performed.
8.6 Sample Recovery. 8.6.1 Preparation. 8.6. I . 1 Roper cleanup procedure begins as soon as the probe is removed from the stack at the end of the
sampling period. Allow the probe to cool. When the probe can be handled safely, wipe off all external particulate matter near the tip of the probe nozzle and place a cap over the tip to prevent losing o r gaining particulate matter. Do not cap the probe tip tightly while the sknpling train is cooling down because this will create a vacuum in the train.
8.6.1.2 Before moving the sample train to the cleanup site, remove the probe from the sample train and cap the open outlet, being careful not to lose any condensate that might be present. Cap the impinger inlet. Remove the umbilical cord from the last impinger and cap the impinger.
8.6.1.3 Transfer the probe and the impingedcondenser assembly to the cleanup area. This area should be clean and protected from the weather to minimize sample contamination or loss.
8.6.1.4 Save a ponion of all washing solutions (toluene/acetonitrile) used for cleanup as a blank. Transfer 200 mL of each solution directly from the wash bottle being used and place each in a separate, prelabeled glass sample container.
8.6. I .5 Inspect the train prior to and during disassembly and note any abnormal conditions. 8.6.2 Sample Containers. 8.6.2.1 Container No. 1. With the aid of an assistant, rinse the probe/nozzle first with toluene and then with
acetonitrile by tilting and rotatins the probe while squining the solvent into the upper end of the probe so that all of the surfaces are wetted with solvent. When using these solvents insure that proper ventilation is available. Let the solvent drain into the container. If particulate is visible, use a Teflon@ brush to loosen/remove the particulate and follow with a second rinse of each solvent. After weighing the contents of the first impinger, add it to container No. 1 along with the toluene and acetonitrile rinses of the impinger. (Acetonitrile will always be the final rinse.) If two liquid layers are present add both to the container. After all components have been collected in the container, seal the container, mark the liquid level on the bonle and add the proper label.
8.6.2.2 Container No. 2. After weighing the contents of the second, third and founh impingers, add them to container No. 2 along with the toluene and acetonitrile rinses of the impingers, the condenser and all connecting glassware. After all components have been collected in the container, seal the container, mark the liquid level on the bonle and add the proper label.
8.6.3 The contents of the fifth and sixth impingers (charcoal and silica gel) can be discarded after they have been weighed.
8.6.4 Sample Preparation for Shipment. Prior to shipment, recheck all sample containers to ensure that the caps are well secured. Seal the lids with Teflon@ tape. Ship all samples upright, packed in ice, using the proper shipping materials as prescribed for hazardous materials.
9.1 Sampling. See EPA Manual 600/4-77-027b for Method 5 quality control. 9.1.1 Field Blanks. Field blanks must be submitted with the samples collected at each sampling site. The
field blanks include the sample bottles containing aliquots of sample recovery solvents, and impinger solutions. At a minimum, one complete sampling train will be assembled in the field staging area, taken to the sampling area, and leak-checked at the beginning and end of the testing (or for the same total number of times as the actual test train). The probe of the blank train shall be heated during the sample test. The train will be recovered as if it were an actual test sample. No gaseous sample will be passed through the sampling train.
9.1.2 Reagent Blanks. An aliquot of t6luene, acetonitrile and the impinger solution will be collected in the field as sepajate samples and returned to the laboratory for analysis to evaluate artifacts that may be observed in the actual samples.
Revision 2 August 1995
10.0 Calibration and S~andardization.
NOTE: Maintain a laboratory log of all calibrations.
10. I Probe Nozzle. Probe nozzles shall be calibrated before their initial use in the field. Using a micrometer, measure the inside diameter of the nozzle to the nearest 0.025 mm (0.001 in.). Make measurements at three separate places across the diameter and obtain the average of the measurements. The difference between the high and low numbers shall not exceed 0.1 mm (0.004 in.). When nozzles become nicked, dented, or corroded, they shall be reshaped, sharpened, and recalibrated before use. Each nozzle shall be permanently and uniquely identified.
10.2 Pitot Tube Assembly. The Type S pitot tube assembly shall be calibrated according to the procedure outlined in Section 4 of promulgated EPA Method 2 (Section 10.1, Reformatted Draft EPA Method 2), or assigned a nominal coefficient of 0.84 if it is not visibly nicked, dented, or corroded and if it meets design and intercomponent spacing specifications.
10.3 Metering System. 10.3.1 Before its initial use in the field, the metering system shall be calibrated according to the procedure
outlined in APTD-0576, Instead of physically adjusting the dry-gas meter dial readings to correspond to the wet-test meter readings, calibration factors may be used to correct the gas meter dial readings mathematically to the proper values. Before calibrating the metering system, it is suggested that a leak-check be conducted. For metering systems having diaphragm pumps, the normal leak-check procedure will not detect leakages within the pump. For these cases the following leak-check procedure is suggested: ~ a k e a 10-min calibration run at 0.00057 &min (0.020 cfin); at the end of the run, take the difference of the measured wet-test and dry-gas meter volumes and divide the difference by 10 to get the leak rate. The leak rate should not exceed 0.00057 d h i n (0.020 cfm).
10.3.2 Afier each field use, the calibration of the metering system shall be checked by performing three calibration runs at a single intermediate orifice setting (based on the previous field test). The vacuum shall be set at the maximum value reached during the test series. To adjust the vacuum, insert a valve between the wet-test meter and the inlet of the metering system. Calculate the average value of the calibration factor. If the calibration has changed by more than 5%, recalibrate the meter over the full range of orifice settings as outlined in APTD-0576.
10.3.3 Leak-check of metering system. That portion of the sampling train from the pump to the orifice meter (see Figure XXXI- I) should be leak-checked prior to initial use and after each shipment. Leakage after the pump will result in less volume being recorded than is actually sampled. Close the main valve on the meter box. Insert a one-hole rubber stopper with rubber tubing attached into the orifice exhaust pipe. Disconnect and vent the low side of the orifice manometer. Close off the low side orifice tap. Pressurize the system to 13-18 cm (5-7 in.) water column by blowing into the rubber tubing. Pinch offthe tubing and observe the manometer for 1 min. A loss of pressure on the manometer indicates a leak in the meter box. Leaks, if present, must be corrected.
NOTE: If the dry-gas-meter coefficient values obtained before and afier a test series differ by >5%, either the test series shall be voided or calculations for test'series shall be performed using whichever meter coefficient value (i.e., before or afier) gives the lower value of total sample volume.
10.4 Probe Heater. The probe-heating system shall be calibrated before its initial use in the field according to the procedure outlined in APTD-0576. Probes constructed according to APTD-058 1 need not be calibrated if the calibration curves in APTD-0576 are used.
10.5 Temperature Sensors. Each thermocouple must be permanently and uniquely marked on the casing; all mercury-in-glass reference thermometers must conform to ASTM E-I 63 specifications. Thermocouples should be calibrated in the laboratory with and without the use of extension leads. If extension leads are used in the field, the thermocouple readings at ambient air temperatures, with and without the extension lead, must be noted and recorded. Correction is necessary if the use of an extension lead produces a change >1.5%.
10.5.1 Dry-gas meter thermocouples. For the thermocouples used to measure the temperature of the gas leaving the impinger train three-point calibration at ice-water, room-air, and boiling-water temperatures is necessary. ~ c c e ~ t the thermocouples only if the readings at all three temperatures agree to *?C (3.6-F) with those of the absolute value of the reference thermometer. .
10.5.2 Probe and stack thermocouples. For the thermocouples used to indicate the probe and stack temperatures, a three-point calibration at ice-water, boiling-u-ater, and hot-oil-bath temperatures must be performed; it is recommended that room-air temperature be added, and that the thermometer and the thermocouple agree to
Revision 2 August 1995
METHOD XXXl
within 1.5% at each of the calibration points. A calibration curve (equation) may be constructed (calculated) and the data extrapolated to cover the entire temperature range suggested by the manufacturer.
10.6 Barometer. Adjust the barometer initially and prior to each test series to agree to within e . 5 mm Hg (0.1 in. Hg) of the mercury barometer or the corrected barometric pressure value reponed by a nearby National Weather Service Station (same altitude above sea level).
10.7 Balance. Calibrate the balance before each test series, using Class-S standard weights; the weights must be within *0.5% of the standards, or the balance must be adjusted to meet these limits.
11.0 Procedures.
1 I .I Sampling Operation. Follow the sampling procedure outlined in Section 8.5. 1 1 2 Analytical. See Method XXX2 - Analysis for lsocyanates by High Performance Liquid
Chromatography (HPLC).
12.0 Data Analjsis and Calcularionr
12.1 Perform Calculations. Round off figures after the final calculation to the correct number of significant figures.
12.2 Nomenclature. A, = Cross-sectional area of nozzle, m' (ftl). B, = Water vapor in the gas stream, proportion by volume. C, = Type S pitot tube coefficient (nominally 0.84 *0.02), dimensionless. 1 = Percent of isokinetic sampling. L, = Individual leakage rate observed during the leak-check conducted prior to the first component
change m3/min (cfm). L, = Maximum acceptable leakage rate for a leak-check, either pre-test or following a component
change; equal to 0.00057 ml/min (0.020 cfm) or 4% of the average sampling rate, whichever is less.
L, = Individual leakage rate observed during the leak-check conducted prior to the 'To component change (i = 1,2, 3 . ~ ) d t m i n (cfm).
L, = Leahage rate observed during the post-test leak-check, m'lmin (cfm). h = Stack-gas dry molecular weight, glg-mole (Ib!lb-mole). M, = ~ o l e c u l a r weight of water, 18.0 &-mole (18.0 Ibllb-mole). P,,, = Barometric pressure at the sampling site, mrn Hg (in. Hg). P, = Absolute stack-gas pressure, mm Hg (in. Hg). P,,, = Standard absolule pressure, 760 mm Hp (29.92 in. Hg). R = Ideal gas constant, 0.06236 mm Hg-dlK-g-mole (21.85 in. Hg-ft'f-R-lb-mole). T, = Absolute average dry-gas meter temperature, K ('R). T, = Absolute average stack-gas temperature, K ('R). Tlld = Standard absolute temperature, 293 K (528 'R). V,, = Total volume of liquid collected in the impingers, and silica gel, mL. V, = Volume of pas sample as measured by dry-gas meter, dscm (dscf). Vm(rtd) = Volume of gas sample measured by the dry-gas meter, corrected to standard conditions, dscm
(dscf). V,,,,, = Volume of water vapor in the gas sample, corrected to standard conditions, scm (scf). V, = Stack-gas velocity, calculated by EPA Method 2, Equation 2-9, using data obtained from EPA
Method 5, m/sec (Wsec). y = Dry-gas-meter calibration factor, dimensionless. AH = Average pressure differential'across the orifice meter, mm YO (in. H,O). p,- = Density of waler, 0.9982 glmL (0.002201.lb/mL). 8 = Total sampling time, min. .
8, = Sampling time interval from the beginning of a run until the first component change, min. 8, = Sampling time interval behveen two successive component changes, beginning with the interval
between the first and second changes, min.
Revision 2 August 1995
METHOD XXXl
eP = Sampling time interval from the final (dh) component change until the end of the sampling mn, min.
13.6 = Specific gravity of mercury. 60 = s e c / m i n . , 100 = Conversion to percent.
12.3 Average Dry-Gas-Meter Temperature and Average Orifice Pressure Drop. See data sheet (Figure XXX 1-2).
12.4 Dry-Gas Volume. Correct the sample measured by the dry-gas meter to standard conditions (20 'C, 760 mm Hg [68 'F, 29.92 in. Hg]) by using Equation XXXl-1:
where:
Eq. XXX 1 - 1
K, = 0.3853 Wmm Hg for mcbic units, ar K, - 17.64 'Win. Hg for English units.
11 should be noted that Equation XXXI-I can be used as written, unless the leakage rate observed during any of the mandator), leak-checks (i.e., the post-test leak-check or leak-checks conducted prior to component changes) exceeds L,. If L, or L, exceeds L,, Equation XXX 1 - 1 must be modified as follows:
a. Case 1 (no component changes made during sampling run): Replace \; in Equation XXXI-1 with the expression:
v, - (L, - L,) 8
b. Case 1 1 (one or more component changes made during the sampling run): Replace V, in Equation X X X 1 - 1 bj, the expression:
and substitute onl)' for those leakage rates (L, or L,) that exceed L,.
12.5 Volume of Water Vapor Condensed.
- where:
K, = 0.001333 m3/rnl for metric units, or K, - 0.04707 ft'lml for English units.
Revision 2 August 1995
METHOD XXXl
12.6 Moisture Content.
NOTE: In saturated or water-droplet-laden gas streams, two calculations of the moisture content of the stack gas shall be made, one fiom the impinger analysis (Equation XXXl-3) and a second fiom the assumption of saturated conditions. The lower of the two values of & shall be considered correct. The procedure for determining the moisture content based upon assumption of saturated conditions is given in the NOTE to Section 1.2 of promulgated EPA Method 4 (Section 4.0 of Reformatted Draft EPA Method 4). For the purposes of this method, the average stack-gas temperature may be used to make this determination, provided that the accuracy of the in-stack temperature sensor is f 1 *C (2 'F).
12.7 Conversion Factors.
EcQE scf
dft3 g/ft3
glftl
12.8 lsokinetic Variation. 12.8.1 Calculation From Raw, Data.
Eq. XXX1-4
where:
K, - 0.003454 mm Hg-ml/rnL-K for maric units,'V K, - 0.002669 in. Hg-R3/mL-'R for English units.
12.8.2 Calculation For Intermediate Values.
where:
K, - 4.320 for metric units, or K, = 0.09450 for English units.
12.8.3 Acceptable Results. If 90% 4 r l lo%, the results are acceptable. If the results are low in comparison with the standard and 1 is beyond the acceptable range, or if l is less than 90%, the Administrator may opt to accept the results.
Revision 2 Augusr 1995
METHOD XXXl
12.9 Concentration of Any Given Isocyanate in the Gaseous Emissions. I. Multiply the concentration of the isocyanate as determined in Method XXX2 by the final
concentration volume, typically 10 mL.
CLocyWtc x sample volume (mL) = amount (pg) of isocyanate in sample Eq. XXX 1-6 where:
- c--,.te - concentration of isocyanate as analyzed by Method XXXZ (pg/rnL).
2. Sum the amount of isocyanate found in all samples associated with a single train.
Total (pg) = Container No. 1 (pg) + Container No. 2 (pg) Eq. XXX 1-7
3. Divide the total pg found by the volume of stack gas sampled (d).
(Total pg) / (train sample volume) = concentration of isocyanate (pglm') Eq. XXX 1-8
13.1 Method Performance Evaluation. Evaluation of analytical procedures for a selected series of compounds must include the sample-preparation procedures and each associated analytical determination. The anal>~ical procedures should be challenged by the test compounds spiked at appropriate levels and carried through the procedures.
13.2 Method Detection Limit. The overall method detection limits (lower and upper) must be determined on a compound-by-compound basis because different compounds may exhibit different collection, retention, and extraction efficiencies as well as the instrumental minimum detection limit (MDL). The method detection limit must be quoted relative to a given sample volume. The upper limits for the method must be determined relative to compound retention volumes (breakthrough). Method Detection Limits may vary due to matrix effects and instrument conditions.
13.3 Method Precision and Bias. The overall method precision and bias must be determined on a compound-by-compound basis at a given concentration level. The method precision value would include a combined var iabi l i~ , due to sampling, sample preparation, and insnumental analysis. The method bias would be dependent upon the collection, retention, and extraction efficiency of the train components. From evaluation studies to date using a djmamic spiking system, acceptable method biases (per EPA Method 301) have been determined for all four isocyanates. A precision of less than 109'0 relative standard deviation (RSD) has been calculated from field test data sets which iesulted from a series of paired. unspiked and spiked trains.
14.0 Pollutio~r Prelmlio~an. Not Applicable.
15. 0 U'oste A,fatanoge~~~e,tr. Not Applicable.
16.0 Rejerences.
1. Martin, R.M., Construction Details of lsokinetic Source-Sampling Equipment. Research Triangle Park, NC, U.S. Environmental Protection Agency, April 197 1, PB-203 060/BE, APTD-058 I, 35 pp.
2. Rom, J.J., hlaintenance, Calibration, and Operation of lsokinetic Source Sampling Equipment, Research Triangle Park, NC, U.S. Environmental protection Agency, March 1972, PB-209 022/BE, APTD-0576,39 pp.
3. Schlickenrieder, L.M., Adams. J.W., and Thrun, K.E., Modified Method 5 Train and Source Assessment Sampling System: Operator's Manual, U S . Environmental Protection Agency, EPA/600/8-89003 (1985).
4. Shigehara, R.T., Adjustments in the EPA Nomograph for Different Pitot Tube Coefficients and Dry Molecular Weights, Stack Sampling News, 2:4-1 I (October 1974).
5. U.S. En\,ironmental Protection Agency, 40 CFR Part 60, Appendix A, Methods 1-5.
Revision 2 August 1995
METHOD XXXl
6. Vollaro, R.F., A Survey o f ~ommercially Available Instrumentation for the Measurement o f Low-Range Gas Velocities, Research Triangle Park, NC, U.S. Environmental Protection Agency, Emissions Measurement Branch, November 1976 (unpublished paper).
17.0 Tables, Diograrns, Fiowchorrs, and Validation Doto. Not Applicable.
Revision 2 August 1995
APPENDIX E
ANALYSIS METHOD FOR ISOCYANATES
MET'HOD XXX2
METHOD X X X 2 - ANALYSIS FOR ISOCYANATES BY HIGH PERFORMANCE LIQUID CHROMATOGRAPHY (HPLC)
1.0 Scope and Application.
I . I Method XXX2 coven the determination of derivatized isocyanates collected by Method XXX1 - Sampling Method for Isocyanates.
Limit of Compound Name CAS No. Quantitation'
2,4-Toluene Diisocyanate (TDI) 584-84-9 0.036 &mL
1,6-Hexamethylene Diisocyanate 822-06-0 0.132 pg/mL (HDI)
Methylene Diphenyl Diisocyanate 101-68-8 0.036 pg/mL (MDU
Methyl Isocyanate (MI) 624-83-9 NAb
'The limit of quantitation is for the instrument conditions given in Section 11 3 .1 at the 95% confidence level for seven replicate injections. bThe limit of quantitation for MI has not been determined.
1.2 This method is restricted to use by, or under the supervision of, analysts experienced in the use of chromatography and in the interpretation of chromatograms. Each analyst must demonstrate the ability to generate acceptable results with this method.
2.1 The impinger contents from Method XXX1 are concentrated to dryness under vacuum, brought to volume with acetonitrile (ACN) and analj.zed by HPLC.
3.0 Dejinirions. Not Applicable.
4.0 Interferences.
4.1 Method interferences may be caused by contaminants in solvents, reagents, glassware, and other sample processing hardlvare. All of these materials must be routinely demonstrated to be free from interferences under conditions of the analysis by preparing and analyzing laboratory method (or reagent) blanks.
4.1 . I Glassware must be cleaned thoroughly before using. The glassware should be washed with laboratory detergent in hot water followed by rinsing with tap water and distilled water. The glassware may be cleaned by baking in a glassware oven at 400 'C for at least one hour. Afier the glassware has cooled, the glassware should be rinsed three times with methylene chloride and three times with acetonitrile. Volumetric glassware should not be heated to 400 'C. Rather, after washing and rins,ing, volumetric glassware may be rinsed with ACN followed by methylene chloride and allowed to dry in air.
4.12 The use of high purity reagents and solvents helps to minimize interference problems in sample analysis.
5.1 The toxicity of each reagent has been precisely defined. Each isocyanate can produce dangerous levels of hydrogen cyanide (HCh'). The exposure to these chemicals must be reduced to the lowest possible level by
Revision 2 August 1995
METHOD XXX2
whatever means available. The laboratory is responsible for maintaining a current awareness file of Occupational Safety and Health Administration (OSHA) regulations regarding safe handling of the chemicals specified in this method. A reference file of material safety data sheets should also be made available to all personnel involved in the chemical analysis. .Additional references to laboratory safety are available.
6.0 Equipment and Supplies.
6.1 Rotary Evaporator. Buchii Model EL-130 or equivalent. 6.2 1000 ml round bottom flask for use with rotary evaporator. 6.3 Separatory Funnel. 500 mL or larger, with Teflon@ Stopcock. 6.4 Glass Funnel. Short stemmed or equivalent. 6.5 Vials. 15 mL capacity with Teflon@ lined caps. 6.6 Class A Volumetric Flasks. 10 mL for bringing sample to volume after concentration. 6.7 Filter Paper. Scientific Products Grade 370 Qualitative or equivalent. 6.8 Buchner Funnel. Porcelain with 100 mm ID or equivalent. 6.9 Erlenmeyer Flask. 500 mL with side arm and vacuum source. 6.10 HPLC with at least a binary pumping system capable of a programmed gradient. 6.1 1 Column. Alltech Altima C18,250 mm x 4.6 mm ID, 5pm particle si te (or equivalent). 6.12 Guard Column. Alltech Hypersil ODs C18, 10 mm x 4.6mm ID, 5pm particle size (or equivalent). 6.13 UV detector at 254 nm. 6.14 Data system for measuring peak areas and retention times.
7.0 Reagenrs and Standards.
7.1 Reagent grade chemjcals should be used in all tests. All reagents shall conform to the specifications of the Committee on Analytical Reagents of the American Chemical Society, where such specifications are available.
7.2 Toluene, GH,CH,. HPLC Grade or equivalent. 7.3 Acetonitrile, CH,CN (ACN). HPLC Grade or equivalent. 7.4 Methylene Chloride, CH:CL,. HPLC Grade or equivalent. 7.5 Hexane, C,H,,. Pesticide Grade or equivalent. 7.6 Water, H,O. HPLC Grade or equivalent. 7.7 Ammonium Acetate, CH,CO,NH,. 7.8 Acetic Acid (glacial), CH,CO,H. 7.9 1-(2-Pyrid!rl)piperazine, (1.2-pp). Aldrich. 99.5+% or equivalent. 7.1 0 Absorption Solution. Prepare a solution of 1-(2-pyridy1)piperazine in toluene at a concentration of
40 mgGOO mL. This solution is used for method blanks and method spikes. 7.1 1 Ammonium Acetate Buffer Solution (AAB). Prepare a solution of ammonium acetate in water at a
concentration of 0.1 M by transferring 7.705 g of ammonium acetate to a 1000 mL volumetric flask and diluting to volume with HPLC Grade water. Adjust pH to 6.2 with glacial acetic acid.
8.0 Saniple Collection, Preservation, Storage, and Transport.
8.1 The components from Method XXX I must be stored at 4 *C between the time of sampling and concentration. Each sample should be extracted and concentrated within 30 days afier collection and analyzed within 30 days afier extraction. The extracted sample must be stored at 4 'C.
9.0 Quality Conmol.
9.1 The correlation coefficient for the calibration curve must be 0.995 or greater. If the correlation coefficienlis less than 0.995, the HPLC system should be examined for problems, and a new calibration curve should be prepared and analyzed.
9.2 A solvent blank should be analyzed daily to verify that the system is not contaminated. 9.3 A calibration standard should be analyzed prior to any samples being analyzed, afier every 10 injections
and at the end of the sample set. Samples must be bracketed by calibration standards that have a response that does
Revision 2 August 1995
METHOD XXX2
not vary by more than 10% of the target value. If the calibration standards are outside the limit, the samples must be reanaly~ed after it is verified that the analytical system is in control.
9.4 A method blank should be prepared and analyzed for every 10 samples concentrated (Section I 1.4). 9.5 A method spike should be prepared and analyzed for every 20 samples. The response for each analyte
should be within 20% of the expected theoretical value of the method spike (Section 1 1.3).
10.0 Calibration and Standardization.
10.1 Establish the retention times for each of the isocyanates of interest using the chromatographic conditions provided in Section 11.5.1. The retention times provided in Table 11.5.1-1 are provided as a guide to relative retention times. Prepare derivatized calibration standards (concentrations expressed in terms of the free isocyanate, Section 12.1) according to the procedure in Section 10.1 .I. Calibrate the chroma?ographic system using the external standard technique (Section 10.1.2)
10.1.1 Preparation of calibration standards. Prepare a 100 pglmL stock solution of the isocyanates of interest from the individual urea as prepared in Sections 1 1.1.1 and 1 1.1.2. This is accomplished by dissolving 1 mg of each urea in 10 mL of ACN. Calibration standards are prepared from this stock solution by making appropriate dilutions of aliquots of the stock into ACN. Calibrate the instrument from 1 to 20 pglmL for HDI, TDI and MDI, and from 1 to 80 pglmL for M1 using at least six calibration points.
10.12 External standard calibration procedure. Analyze each derivatized calibration standard using the chromatographic conditions listed in Section 1 1.5.1 and tabulate peak area against concentration injected.
The working calibration curve must be verified on each working day by the measurement of one or more calibration standards. If the response for any analyte varies from the target response by more than lo%, the test must be repeated using a fresh calibration standard(s) after it is verified that the analytical system is under control. Alternatively, a new calibration curve may be prepared for that compound.
11.0 Procedures.
I 1 . I Preparation of isocyanate derivatives. 1 1 -1.1 HDI, TDI, MDI. 1 1.1.1.1 Dissolve 500 mg of each isocyanate in individual 100 mL aliquots of MeCj, except for MDI
which requires 250 mL of MeCh. Transfer a 5 mL aliquot of 1,2-pp (see Section 7.10) to each of the solutions, stir and allow to stand overnight at room temperature. Transfer 150 mL aliquots of hexane to each of the solutions to precipitate the isocyanate-urea. Using a Buchner funnel, vacuum filter the solid-urea and wash with 50 mL of hexane. Dissolve the precipitate in a minimum aliquot of MeCb. Repeat the hexane precipitation and filtration .m,ice. After the third filtration, dry the crystals at 50 'C and transfer to bottles for storage. The crystals are stable for at least 21 months when stored at room temperature in a closed container.
Table 11.1.1-l
hlolecular Weight of the Free Isocyanates and the Isocyanate-Urea
Analyte M\V (Free Isocyanate) MW (Derivative)
1,6-HDI ' 168 494.44
MDI 250.25 576.65 -
Revision 2 August 1995
METHOD XXX2
K = 35.3 1 ft 'lrn if V,, is expressed in Englii units. = 1.00 m3/m3 if VMM is #pnsd in metric units.
V,,, = Volume of gas sample as mcuured by a Qy gu mctm; conected to ct~dard conditions, dscm (dscf).
13.0 Method Perfarmanee.
13.1 Method detection limit (MDL) concentrations were obtained by determining the standard deviation of the area count for seven replicate injections and then multiplying the standard deviation for seven replicate injections by the student's t-valve at the 95% confidence level (3.143).
13.2 This method has been tested for linearity over the range fiom 2 x MDL to 500 x MDL.
14.0 Pollurion Prevention. Not Applicable.
15.0 Jl'aste hhagement . No1 Applicable.
16.0 References. Not Applicable.
17.0 Tables, Diagrams, Flori~liarts, arid b'alidation Dota. Not Applicable.
Revision 2 XXXZ -6- August 1995
1. NTlS ACCI
PB95-- - 1 ; 3. Oocurnent Title >
: Laboratory ~evsbpment and Field Evaluation of a
i Generlc Method for Sampllng and Analysls of Isocyanate:
COMPUTER PROOUCT
E
Report No. 6OO/R-95/144 NERL-RTP 0-20 5 .
(CUT HERE AND FILE BY TITLE)
PLEASE ENSURE YOUR CORRECT RETURN ADDRESS IS ENTERED ON THE FRONT OF THIS CA INSTRUCTIONS:
F~ll ~n boxes 2, 3, 4, 5, 6, and 7 A returned card ~nd~cates your rnatend NTlS will fill ~n boxes 1 and 8 and msl to addressee 1s be~ng processed mto the NTlS system. Staple card to front top cover of document anbn~ng For add~bonal cards, wnte Inlormatron the Reporl Documenlat~on Page Services Branch, NTIS.
U S DEPARTMENT OF COMMERCE NATIONAL TECHNICAL INFORMATlON SERVICE SPRINGFIELD VA 22161
4 Source C l m t Code EPA/ORD
5 Procedure No ( I f ass~gned)
2 12 6 No coples sent
1 Planned Press Release
' 7 CONTACT Name/ Phone Number
. *
P.O. Frank Wllshire 8 NTIS DOMESTIC PRICE CODES AND CURRENT $ VALUES
L, 6k .
Top Related